Mividas Installer / VM

[2.0.1] - 2026-10-27

A maintenance release for 2.0.0 following extensive security- and functional reviews. It adds extended audit log of Installer changes and sign-ins, and fixes single sign-on, VM-upgrades, backup among other things. Hosts running STIG or FIPS get further hardening, and the release includes the latest security updates and fixes.

Added

• Add status snapshots to the log view – the server's periodic snapshots of service states, memory and disk use appear next to the system log.
• Add an Installer audit log – every change, sign-in, failed sign-in and sign-out is logged with the user, the client address and the fields submitted, never passwords, secrets or personal data.
• Use local images while the registry is unreachable – on a server that cannot reach the Mividas registry, for example behind a firewall, pulls use the images already on the server at once, and the registry is used again once it answers.
• Let the console wizard lock its settings – it shows whether the console can change settings and offers to lock them from the next reboot, after which it points to sudo onboard_wizard.

Fixed

Sign-in

• Fix the sign-in lockout and the client address behind an external load balancer – after ten failed sign-ins the Installer locks out only the client that failed, not every user, and when its web interface is served through a load balancer listed under "Trusted HTTP load balancer/reverse proxy IPs" in Server settings, the lockout, the logs and the audit log use the real client address.
• Keep Installer sign-in working while its cache is down – signing in, including single sign-on, works while the Installer's cache service is stopped or restarting, where it failed with "Unable to create a new session key"; the upgrade signs everyone out once.
• Fix SAML sign-in and sign-out with saved or unreachable metadata – SP metadata saved in the SAML settings, including offline SP metadata XML, is used by the Installer and by Core, Meeting Portal and Identify, the last good identity provider metadata is kept, an identity provider whose metadata cannot be fetched gives an error page rather than a server error, one that sends an empty attribute such as an e-mail address can sign users in, and sign-out works in these cases too and with providers such as Shibboleth.
• Support SAML identity providers that accept only POST – sign-in works with identity providers that offer no redirect binding, and an assertion the Installer cannot decrypt, such as one from Keycloak encrypted with RSA-OAEP and, shows the sign-in error page rather than a server error
• Fix slow LDAP tests and LDAP group checks – on a domain that lists many LDAP servers in DNS, the LDAP test and saving LDAP settings try at most three of them rather than waiting on each, and where the Installer's own sign-in uses LDAP, OpenLDAP and 389 Directory Server group memberships are found.

Certificates, licenses and offline bundles

• Fix certificate chains and "Delete all unused" – a certificate uploaded without its root passes the check and is served, a CA or intermediate uploaded after the certificate it issued joins the served chain at the next deploy, a certificate file left incomplete by an interrupted write is repaired, "Delete all unused" keeps every key pair a setting uses, such as for SAML, OIDC or OpenLDAP, and editing a CA, re-uploading a CA bundle, Ed25519 certificate requests and reusing a SAML or OIDC certificate work.
• Apply CA trust changes without a restart – trusting or removing a CA, and turning revocation checks on or off, take effect at once for all of the Installer's connections, programs on the server find every trusted CA, and running products and services see trust changes and refreshed revocation lists.
• Fix revocation list refresh and TLS 1.1 for older clients – the refresh succeeds when some distribution points cannot be reached, for example from behind a firewall, certificate chains with an intermediate CA refresh without reporting a failure, and the TLS setting for older clients accepts TLS 1.1.
• Fix offline bundle imports and offline deploys – uncompressed bundles import, a product deploys when its database image is not on the server, other missing images stop the deploy cleanly, and disk cleanup keeps the images upgrades need.
• Fix license activation by order number – activation completes when the license server issues a new key.
• Apply offline license files fully – the products, features and settings in an uploaded license file take effect, also for files uploaded before this release.

Deploys and Installer upgrades

• Fix starting a new server in air gapped environments – a VM deployed from the image for the first time brings up the Installer and its database with no connection to the registry.
• Fix Installer upgrades that stall or do not roll back – an upgrade keeps going while Docker is busy, without pauses of up to several minutes, survives a restart of the Installer's database, returns to the previous version when its retry also fails, starts on Debian 13 when run from the server itself, and gives up on a registry that stops answering, so later Installer actions are not held up.
• Let Installers older than 1.8.18 upgrade – the upgrade reaches Docker where it timed out, and repairs the file permissions the older Installer, which ran as root, left behind, where the new version stopped with "Permission denied".
• Keep a beta or custom Installer version when its services are recreated – after a repair or another step that recreates the Installer's services, it starts again at the version it ran, also offline, rather than at the stable release.
• Share memory between products on one server – products installed together are each sized to their share of the server's memory from the first install and held to it, so a redeploy keeps every product's services running.
• Fix deploys that stop on a missing image, a service update or upload permissions – a deploy completes when a service the product does not use, such as Core's proxy, has no image on the server, recovers when a service cannot be updated in place, and repairs or reports upload directories the product cannot write to, as after a restore.
• Fix deploys started from the web UI – the new version and its settings are recorded when the deploy finishes, also if the browser tab was closed or the connection dropped before then.
• Warn about products installed without a Core – the product page warns when Meeting Portal, LDAP admin or Connect has no Core in the same product to connect to, which leaves its Booking API address empty, and the progress console clears its "No output" warning once output arrives.
• Fix e-mail through mail servers on port 465 and without a sender address – with SMTP+SSL selected, Core, Meeting Portal and Identify use TLS from the first byte, as servers on port 465 expect, where their mail timed out, and without a sender address they use their own default sender. SMTP+SSL settings on ports other than 465 are switched to SMTP+TLS (STARTTLS), which is what they already used.
• Keep $ in database and SMTP passwords – a password containing $ reaches Core, Meeting Portal and Identify as entered, for an external database and for the mail server.
• Support Redis Sentinel as the external Redis – with an external Redis (beta) set to the "Redis sentinel" protocol, Sentinels entered as host:port with a password or none, database 0, and a master named mymaster, the settings save and Core 4.2, Meeting Portal 2.12, Identify 1.2 and later start; for earlier versions the save is refused.
• Apply Meeting Portal's "Disable dynamic customer shared key" – when ticked, which is the default, bookings use the default customer shared key. Installations that kept the default switch to it at their next deploy; untick it to keep a key per customer.
• Apply Installer changes on save – saved Server settings and load balancer changes, and the end of an Installer upgrade, reach the server at once rather than at its next scheduled pass, and an upgrade started from the web page keeps the Installer available rather than answering with error 503.

Logs and remote syslog

• Fix remote syslog over TLS and the severity of forwarded log lines – forwarding over TLS works on every host, with the system logger and its TLS support installed where missing, the console wizard says when it cannot forward over TLS, and log lines from the products and the Installer reach the remote server at the severity they were logged with, not all as info or error.
• Fix the log view's decoding and redaction – a log that cannot be decoded leaves the other logs visible, the MAC address, IPv6 address, domain and instance name redactions each work on their own without garbling the output, and console errors name the reason.
• Fix remote audit, audit archives and journal limits – upgraded hosts can send audit records to a remote server, also offline, forwarding resumes by itself after the collector was unreachable, a collector address or port the forwarder cannot use is refused, every audit archive is kept when the disk is full, the audit rules load at boot, and journal size limits apply immediately.

Network and proxy

• Fix proxy exceptions and network test timeouts – address ranges and IPv6 addresses in proxy exceptions are honoured, and the network test's TLS and HTTP checks stop after a time limit.
• Save console wizard network settings safely – a full disk or power cut while saving leaves the previous settings in place, an MTU below 68 is refused, and Debian 11 hosts on DHCP get their address with an IPv6 DNS server set.
• Keep DNS and NTP settings on hosts from 2.0.0 images – the console wizard's DNS servers and the short DNS timeout survive reboots and lease renewals, and the NTP server DHCP hands out is used.
• Restore network isolation faster – isolation returns sooner after a container restarts, holds while a network's rules are rebuilt, and fail2ban's SSH bans stay in force on Debian 11 hosts.

Databases, backups and restore

• Keep a product's database version on redeploy – a product whose database was set up before 2.0.0 keeps its database image when redeployed, also offline, so the move to Debian 13 and the index rebuild that must go with it happen only in the database upgrade.
• Protect data during PostgreSQL upgrades – the previous database is kept and nothing deleted if the restore reports errors, a stopped upgrade brings the product back for a retry, and an upgrade interrupted after its data was converted keeps the new version and offers the index rebuild again.
• Recover from failed database restores – a backup is checked before anything is stopped, the failure is reported and the original database and product are brought back, a restore stops only that product, and a forced restore starts empty.
• Restore product databases in a full restore – the products are redeployed once the Installer reaches its database, so their databases are restored too, a full backup of Core holds both its databases, and a partial backup says what it leaves out.
• Run freeze steps for quiesced snapshots – quiesced snapshots on KVM and Proxmox run the appliance's freeze steps, and on every platform those steps give up after five minutes so the snapshot goes ahead.

Upgrades to Debian 13 and package updates

• Keep Debian 13 servers booting – an upgraded server boots on every restart even where updating the boot loader failed, and one already affected is repaired before its next restart.
• Keep UEFI hosts on the updated boot loader – shim and GRUB security updates also reach the boot entry a UEFI host starts from.
• Fix the boot loader and kernel in the Debian 13 upgrade – BIOS hosts, also with several disks, keep a working boot loader, whose install is retried until it succeeds, boot loader settings an operator added are kept, hosts with a GRUB password start without a prompt after boot loader updates, a missing kernel is reinstalled, and the upgrade completes on hosts that had no separate /tmp.
• Fix the Debian 13 upgrade from Debian 10 – the offline bundle upgrades Debian 10 hosts too, and the upgrade stops if a package in the bundle is damaged.
• Switch to Debian's container runtime on Debian 13 – on hosts installed with Docker CE, the upgrade replaces it with Debian's runtime, online and offline, and the restart afterwards is not held up.
• Clean up after the Debian 13 upgrade – packages only the previous release used, its offline bundle (about 250 MB) and, once the new kernel runs, the previous kernel are removed.
• Show uptime and finish interrupted OS upgrades – the dashboard shows uptime and offers to finish an upgrade that stopped part way, also after a power cut or an unexpected reboot, or reboot onto the installed kernel, showing "Reboot needed".
• Queue Installer actions while the server is busy – actions run in turn and once each, a running OS upgrade is left alone when started again or reloaded, a security update and an OS upgrade never run at the same time, and Upgrade OS says when the server could not take the request.
• Fix upgrades from 1.8.x and time synchronisation – an upgrade from 1.8.x completes in one pass rather than ending with a syntax error and resuming later, and hosts updated to 2.0.0 from 1.8.17 or earlier, which were left without time synchronisation, get chrony with their previous NTP servers.
• Repair partly installed dbus and lnav – on hosts first installed with 1.8.9 to 1.8.13, the next update repairs them so security updates and chrony install again, and the dashboard says when an administrator must decide.

Console wizard

• Fix the console wizard's admin password prompt – the prompt appears on screen in initial setup, in Security settings and when FIPS asks for a new password, and the admin account stays disabled until the new password is stored.
• Keep current values in console wizard forms – network, syslog and SNMP forms show the current settings, static routes survive a switch to DHCP, and turning remote syslog off stops forwarding.
• Enforce disabled SSH password login – the console wizard's setting holds even when another SSH setting, such as cloud-init's, turns it on, the wizard shows the setting sshd uses, and it asks first when no account has an SSH key the server accepts (with STIG on, Ed25519 keys do not count).
• Fix host name changes – lines other tools add to /etc/hosts take effect after a rename.
• Improve console wizard disk cleanup and status – cleanup removes old debug logs and, while console changes are enabled, asks before removing every unused image, and the status screen stays current and fits 80×24 and serial consoles.

Docker, swarm and host maintenance

• Fix user namespace remapping (beta) – the Installer starts on the first remapped boot, turning remapping on or off keeps data, owners and permissions, a failed or interrupted change is undone with Docker stopped, and TLS keys, settings and logs stay closed to other accounts.
• Keep swarm networks clear of customer networks – recovery from expired swarm certificates keeps overlays in 100.106.0.0/16, outside 10.0.0.0/8, and restores the clock and time synchronisation.
• Restart products after the swarm is recreated – once the swarm is recreated as its recovery steps advise, the reverse proxy and the products start again.
• Fix moving /home to a separate data disk – cli os enable-data-split works on every image, refuses a disk that is not blank, copies only /home's own files, and keeps Docker stopped until the reboot when /home is in use.
• Fix cli component exec and cli component logs – exec returns when the command finishes, and logs runs without extra arguments.
• Log only maintenance changes – the periodic pass logs only what it changed, and restarts the audit service only when its settings change on hosts that send audit records to a remote server.
• Fix SSH key handling for accounts – a key added to an account always starts on its own line, cli user add checks a key before creating the account, and adding, removing or listing a key never follows a link in the operator's home.

Security

Sign-in

• Fix authentication bypass through SAML response replay (SAML sign-in only) – a captured SAML response cannot sign anyone in again or in another browser, sign-out messages must be signed for the session they end, oversized messages are refused and identity provider metadata is fetched only over HTTPS.
• Fix potential takeover of local Installer accounts through directory sign-in – SAML or LDAP sign-in never takes over one of the Installer's local accounts, and directory users outside the required user group are refused; only installations with SAML sign-in, or with LDAP set up for the Installer's own sign-in, are affected.
• Fix potential privilege escalation through crafted group names – group names from the identity provider and the directory are matched whole, so a group whose name contains a comma cannot match an administrator group; this needed a user who can create or name their own groups in the identity provider or directory, where groups are sent as distinguished names or comma-separated lists.
• Harden the first Installer password – the password chosen when the Installer is first opened needs at least twelve characters; FIPS and STIG hosts keep asking for fifteen across all four character classes.
• Fix potential cross-site request forgery – actions such as restarts and license refreshes run only from the Installer's forms, never from a plain link, and an upgrade takes its target version from the submitted form only.
• Harden Installer sessions – the session cookie is Secure with a 40-character mixed-case key, and the web server hides its software.
• Fix potential credential disclosure for the LDAP service account (LDAP only) – the LDAP test sends the stored password only over an encrypted connection, with a server-log warning when it falls back to an unencrypted one, connections stop when encryption fails unless verification is off, referrals are followed only when turned on and certificates are not verified, and a directory server found through DNS must belong to the configured domain.

Certificates, licenses and signing

• Fix the FIPS 140-3 cipher setting on hosts without FIPS mode – with that setting chosen on a host where FIPS mode is off, HTTPS clients connect with approved TLS 1.2 suites only, while FIPS hosts accept TLS 1.2 and 1.3.
• Fix the internal certificate authority's expiry – the Installer renews it before it expires, and certificates it issues never outlive it.
• Fix potential addition of untrusted certificate authorities through revocation lists – a revocation list is used only with a valid signature, also on Debian 11, only the verified list itself is added to the host's trust, an expired list or one older than the installed copy never replaces it, and the reverse proxy reads the Installer's certificates read-only; this needed an attacker on the network path to a revocation list published over plain HTTP.
• Fix license signature verification – uploaded license files are checked against their signature, and license-server responses dated more than an hour from the host's clock are refused.
• Fix potential tampering with offline bundles – bundles are signed over their whole content, importable for 60 days, carry fresh license information and leave out registry credentials and key material; only installations that use offline bundles are affected.
• Harden Installer upgrades – an upgrade accepts only a well-formed version tag.

Deploys and the Installer image

• Harden the stored product connection settings – the copy the Installer's database keeps of each product's generated database, Redis and embedded LDAP settings, passwords included, is encrypted like the password fields.
• Harden product deploys – the product deploy definition settings file is removed after each deploy.
• Harden product deploys against host mounts – products cannot mount system directories such as /etc, /root, /home or the whole file system, only appliance data and what each service needs.
• Fix potential disclosure of an external database's superuser login – when a product's own login cannot be created the deploy stops, instead of giving the product the server's superuser login, and an existing database is taken over only when it is empty and owned by the product's own login; only external PostgreSQL servers with "Create new user/database automatically for each component" turned on are affected.
• Harden passwords during deploys and restores – passwords stay off command lines, where other processes on the host could read them, and out of restore output.
• Fix potential stored cross-site scripting in license agreements – HTML in end-user license agreements is shown as text, and an acceptance record keeps only the accepting person's browser and address.
• Harden the Installer image – it ships without developer tools, its production settings cannot switch to debug or test mode, and its temporary directory is shared safely between users.
• Harden the Installer's own containers – its setup container runs without extended privileges, and every Installer service has a process limit.
• Harden the bundled service images – the reverse proxy, database, cache, connection pool and socket proxy images contain no setuid or setgid programs, and the reverse proxy image has no shell or package manager.

Host, accounts and audit

• Fix potential command injection in the console wizard (console access required) – every value is checked before it is applied, so DNS servers and gateways must be IP addresses, routes valid networks, proxies http or https addresses and a pre-hashed GRUB password well-formed, proxy passwords and SNMPv3 secrets stay hidden, and the SNMPv3 passphrases and the GRUB password hash are kept readable by root only.
• Fix regression since sysctl handling change in Debian 13 – network hardening settings such as disabled IPv6 router advertisements apply again after every reboot, and each VM installed from the image gets its own DHCP identity.
• Harden root maintenance tasks – the tasks that run as root on the server treat every file the Installer can change as untrusted, so a link, a setuid file, a special file or a staged package placed there cannot make them act on other files, stall them or run code as root; the temporary directory keeps its sticky bit, and core dumps are disabled on every host.
• Harden auditing – kernel module loads, sudoedit, cron changes and Debian 13's login records are audited on every host from the next reboot.
• Harden logins and the audit logs – a failed password login over SSH, the console, su or sudo waits four seconds on every host, and only root can read the audit records.
• Harden passwords on every host – new passwords for every account other than root are checked for strength whatever the host's hardening.

Security updates

• Fix security updates that were not installed – the security updates in the offline bundle are installed on every update, as are Debian's on Debian 11 hosts without a bundle, and those from Debian's security archive where unattended upgrades are on.
• Harden the package signing key – the Mividas key is trusted only for the Mividas package repository, and the server reports when an update bundle brings a different key.
• Update the VM, the Installer and the reverse proxy with the latest security fixes – the VM runs the latest Debian 13 kernel and package updates, and the reverse proxy the latest Traefik 2.11 release.
• Upgrade VM packages: linux-image-cloud-amd64, hyperv-daemons, qemu-guest-agent, perl, libglib2.0, libpcre2, openssl, libssh2, python3.13, libsqlite3, gzip, glibc, libexpat1, curl, libbrotli1, dhcpcd-base, libpng16
• Upgrade the Installer container: urllib3, certifi, openssl, libpcre2, libsqlite3, gzip, glibc and container-helper
• Upgrade service images: the reverse proxy (Traefik, rebuilt with stdlib updates), and Redis, PgBouncer, PostgreSQL and the placeholder service with Debian, Alpine and Go updates
• Related CVEs: CVE-2026-56853, CVE-2026-56860, CVE-2026-56862, CVE-2026-71324, CVE-2026-78669, CVE-2026-80521, CVE-2026-80837, CVE-2026-80841, CVE-2026-80849, CVE-2026-80850, CVE-2026-88009, CVE-2026-89761, CVE-2026-97417, CVE-2026-97602, CVE-2026-97687, CVE-2026-97689, CVE-2026-98222, CVE-2026-98260, CVE-2026-98980, CVE-2026-98982, CVE-2026-99999, CVE-2026-111943
• Upgrade the OpenSSL FIPS module on FIPS hosts: 3.5.4 -> 3.5.8

Changed

Installer

• Explain open SAML claims in help text – the SAML settings say that an empty group or role claim admits every authenticated user, and that Installer logins are for trusted operators only.
• Limit single sign-on and LDAP sessions – on hosts without STIG or FIPS, Installer sessions started through single sign-on or LDAP end 12 hours after sign-in regardless of activity, and signing in again restarts the clock.
• Apply the TLS and revocation settings to the Installer itself – its own connections, such as to the license server and the registry, check revocation when that is turned on, its web address honours the TLS 1.3 and FIPS cipher settings, and it connects to an external PostgreSQL database with TLS, so a database server without TLS needs its TLS mode set to disable or prefer.
• Report failed update steps – the Installer update log shows a failed command's last lines of output, and cli commands and the deploy command end with an error status.
• Confirm Installer upgrades from the console wizard – the wizard asks before it upgrades the Installer, runs the same upgrade as cli installer upgrade and says whether it finished.
• Keep local backups across reboots and readable by root only – local backups and database dumps are stored in /opt/mividas/backups, which a reboot leaves in place, every backup archive is readable by root only from its first byte, and a full backup stops when a database dump fails unless --allow-partial is given.

Host and console wizard

• Close SSH sessions whose client has stopped responding – on servers installed from this release's image, an SSH session whose client no longer answers is closed after about ten minutes.
• Remove any manually added operator user accounts from the docker group, require setting a personal sudo password - cli user add prints a one-time password, and updates name the accounts that still need cli user reset-password.
• Use the platform's network settings on a new VM until the console wizard sets them – a VM on a platform that supplies its address and routes, for example through cloud-init, uses them, or DHCP where none are supplied, and keeps the wizard's settings once they are made.
• Hold back security updates that would remove packages – when an update, for example from an outdated offline bundle, would remove installed packages, cli apply names them and ends with an error; installing a current offline bundle resolves it.

STIG and FIPS – Added

• Offer key-only SSH login on STIG and FIPS hosts – turning on STIG or FIPS in the console wizard offers to make SSH key-only once an account has a key it can use.
• Warn about local sign-in on STIG hosts – the product page names components whose users can still sign in with local accounts.
• Send remote syslog over TLS with FIPS-approved algorithms on Debian 13 FIPS hosts – remote syslog over TLS, and the other system programs built on the GnuTLS library, use FIPS-approved algorithms only.

STIG and FIPS – Fixed

• Name only installed components in the deploy reminder – after FIPS or STIG changes, the "Deploy changes" reminder lists only the components that are installed.
• Improve database system migration for FIPS hosts – databases start with FIPS enabled, passwords stored before FIPS was enabled are migrated, products installed before 2.0.0 keep reaching their databases, and the Installer stays connected to its own database while database logins move to the FIPS-approved SCRAM method.
• Fix turning FIPS on and off – it can be re-enabled before the disabling reboot or disabled when set by hand, and turning off "Strip bundled CAs" restores every root CA.
• Fix the Debian 11 FIPS notice – the console wizard's notice that the OpenSSL FIPS provider needs Debian 13 names the cli distupgrade command that upgrades the host.
• Let passwords be set with the CAT III profile on STIG hosts – with STIG and its CAT III profile on, passwords can be set and changed again.

STIG and FIPS – Changed

Sign-in, SSH and audit

• Limit Installer sessions to 8 hours on STIG and FIPS hosts – every Installer session, local sign-ins included, ends 8 hours after sign-in regardless of activity.
• Lock accounts after failed logins on STIG hosts – an account is locked after three failed logins, and SSH sessions whose client stops responding are closed (on every host installed from this release's image, see "Close SSH sessions whose client has stopped responding").
• Apply password ageing to existing accounts on STIG hosts – accounts with a password get the 60-day maximum and 1-day minimum age, and SSH key users are never locked out by it.
• Apply the password strength rules to root on STIG hosts – the rules also apply when root sets a password, including the console wizard's admin password.
• Store passwords with on STIG and FIPS hosts – new passwords use whenever STIG or FIPS is on, and the console wizard asks for a new administrator password when needed
• Use FIPS-approved SSH on STIG-only hosts – SSH uses approved algorithms and accepts only RSA and ECDSA keys, accounts left without one are named at login and by cli apply, new certificates cannot use Ed25519, and enabling FIPS or STIG keeps SSH access.
• Keep a separate sudo log on STIG hosts – sudo writes its own log, readable by root only and rotated weekly.

Cryptography

• Clarify FIPS ciphers as approved – the Installer and the console wizard call them FIPS-approved, and the wizard says the OpenSSL provider has no certificate of its own.
• Use the Go Cryptographic Module in the reverse proxy on FIPS hosts – the traefik reverse proxy uses the certified Go module (v1.0.0-c2097c7c) in FIPS mode.
• Regenerate the OpenSSL FIPS configuration on the host – it is used only if it matches the build, so a changed module leaves the host out of FIPS provider mode with a warning and keeps containers from starting, and a module that fails its self-test leaves the standard configuration in place, so SSH, package updates and the Installer's database keep working; the module is OpenSSL 3.5.8, the version submitted for FIPS 140-3.

[2.0.0] - 2026-09-12

The appliance moves to Debian 13, the largest platform update the product has had.

It is a new foundation for the whole system: a 6.12 long-term-support kernel carrying years of scheduler, memory-management and I/O work, Python 3.13, OpenSSH 10, and an OpenSSL 3 whose provider architecture is what modern cryptographic policy is built on. Years of accumulated security fixes arrive in a single step, and the platform is supported into the next decade.

Existing installations upgrade in place, from the web interface, keeping their data, machine identity and certificates — no reinstall and no migration.

Added

• Upgrade the appliance to Debian 13 from the dashboard. Hosts on Debian 10, 11 and 12 are all carried through, rebooting as often as they need to, with the progress, the log and the final outcome in the web console — and a notice that says up front that the server reboots itself and cannot be put back without a backup or snapshot
• STIG hardening and FIPS provider mode are listed under Security below. Both require a separate license — contact us for more information

Security

STIG and FIPS provider mode

• Warn at login when a host is set up for FIPS provider mode but is not actually running it, and repair it on the next configuration pass by returning to the standard cryptographic configuration, so the host can be logged into again. A host in that state fails every cryptographic call, so it stops accepting new SSH sessions and cannot be logged into to be repaired by hand
• FIPS provider mode is delivered as a package carrying its own module, so nothing has to re-seal a module the distribution owns. It is installed on Debian 13 only, and a host that is upgraded rather than rebuilt picks it up on its next package run
• Passwords set in the Installer, including the one chosen when it is first opened, follow the DISA baseline on a host reporting FIPS or STIG: fifteen characters across all four character classes. Elsewhere the first password meets the same requirements as any other
• Ed25519 and Ed448 keys are refused under FIPS. Both are approved algorithms there, but neither CNSA suite lists either curve and DoD PKI issues no certificate on one, so a key generated on one is unusable to the deployments that ask about FIPS in the first place
• Remind operators in the console wizard's compliance menu that the STIG login banner is theirs to supply, and name the files to edit for SSH and for console logins
• The STIG login banner is left to each deployment, since the required wording differs. The file ships empty with SSH already pointed at it — supply your organisation's own text

[1.8.21] - 2026-08-27

Added

• Run more than one copy of a component's background and web services, each sized from the memory the Installer allocated to that component rather than from a fixed figure in its definition. A component has to ask for this, so existing installations are unaffected
• Show real progress for long-running operations in the web console: numbered steps and a progress bar during a deploy, an upload progress bar and an immediate acknowledgement when importing an offline bundle, and a clear warning when a run was cut short rather than a partial log that looks like a finished one
• Show "Waiting for the database" or "Upgrade in progress" while the database is still starting or migrations are running during an upgrade, instead of a server error page. The pages refresh themselves, so a tab left open recovers on its own
• Keep the "settings changed — needs deploy" reminder until the component is actually deployed. It is now recorded per component, so it survives logout, appears in other browsers, names which components are waiting, and stays up if a deploy fails
• Replace a VM's operating system disk while keeping its data, machine identity and SSH host keys: attach a disk written from a normal image, boot it, and it adopts the existing installation instead of coming up as a stranger on empty storage. New cli os commands cover the status, relabel and adopt steps, and can create the separate data disk on a VM that does not have one
• Collect docker service and host status every 15 minutes into the existing log collection, so a support bundle gathered after an incident shows what was running when the problem happened — including previous task states and the error a rescheduled task died with — not only what is running now
• Add cli traefik prune-certs to clear unloadable certificates from the load balancer and Installer proxy TLS configuration
• Report the host's time synchronisation and warn when it has none: the console wizard's status view shows the configured time sources and whether the clock is actually set, and the dashboard warns when the host has no time source at all. A clock that has quietly stopped being set breaks TLS in ways that look like anything but a clock
• Show what the Installer is running on beside the active license: CPUs, memory, uptime, the Installer version, and the host operating system, kernel and container runtime. Each row appears only where the host can answer for it, so the page still renders when the container runtime cannot be reached
• Upgrade the PostgreSQL databases of managed components, and the Installer's own, to version 16. The upgrade runs from the maintenance page with a confirmation step, and is offered only where the running version and the component's image make it applicable
• Turn on certificate revocation checking for outbound TLS, with the coverage in view: revocation lists are now collected for intermediate authorities as well as roots, and the Installer reports how many issuers are covered, uncovered and failing. It is off by default, because a connection is refused outright when an issuer has no current list
• Choose where audit records go: kept on the host, sent to the system log, or sent to an audit collector. The host's own copy is the default, since the other two cost something on every event and neither is worth paying for by accident
• Check that a SAML identity provider's metadata address is reachable when it is saved. The Installer's own SAML refuses the save, since a mistake there locks the operator out of the server they are configuring; a component's SAML records a warning instead, because the provider is as likely as anything else to be undeployed when the form is filled in
• Check the strength of the password chosen when the Installer is first opened, and list the requirements above the field
• Stream an Installer backup straight over SSH: cli installer backup writes the archive to standard output and cli installer restore reads it back, matching the other backup commands
• Report a run whose output has stopped arriving as stopped, rather than leaving the console spinning. A dropped connection and a slow step look the same from inside the page, so a run that had already ended went on looking active
• Offer the host's own name back in the console wizard's hostname prompt rather than opening empty, where an operator confirming the name they already had had to retype it in full
• Report the deployed message broker version, and tell when a previous broker has been drained and can be shut down. A broker that cannot be reached is never reported drained, since stopping it then would discard whatever it still holds

Fixed

• Fix an Installer upgrade appearing to hang and then ending on "Giving up". It now waits long enough for the host to pick the request up, reads the host's upgrade log from a path it can see, and reports an error rather than spinning once it does time out
• Fix a deploy looking stalled because an internal proxy container was removed mid-deploy, which left the web interface unable to show progress until the stack recreated it
• Fix "value too long" errors on installations that were upgraded through 1.8.12 or 1.8.13, where some encrypted database columns were left at their original width. A migration widens them; fresh and healthy installations are untouched
• Keep the Installer usable when its database connection pool is down, by connecting to the database directly and returning to the pool automatically once it is back
• Fix a Portal or Core connection test reporting success when the far end had rejected it. A rejected API key, a host that does not answer the booking API, and other errors are now reported separately
• Fix a change to the Installer's time zone not taking effect until it was restarted
• Fix container names not matching, and deploy status output going missing, on installations running plain compose rather than swarm
• Skip the post-deploy database tuning step for the Outlook Add-in, which has no database of its own
• Stop trying to fetch an image that offline installations cannot reach, which only filled the log with errors
• Fix a package installation looping while holding the package manager's lock, which left the VM needing a reboot before any further upgrade or configuration change would run. Package operations are now fully non-interactive, keep existing configuration files, never start on a package the offline bundle does not carry, finish an interrupted transaction automatically, and release a lock held by a dead process after an hour
• Fix a TLS certificate that cannot be loaded taking HTTPS down altogether, including the Installer's own interface on :8999 — which is where an operator would go to repair the certificate. The broken entry is dropped so the load balancer falls back to its built-in certificate, and it returns once the certificate is fixed
• Fix the console wizard restarting in a loop on a VM that has no serial port, which on a hardened host was enough on its own to overflow the kernel's audit queue and leave the machine unresponsive
• Fix SSH refusing to start after an upgrade — resetting every connection, so nobody could log in to repair it — on hosts whose configuration named a host key or key exchange algorithm the newer SSH no longer recognises. The names are now checked against the running daemon rather than a fixed list, and only the ones it rejects are removed
• Fix a forced log rotation destroying retained audit records on hosts that keep them, and the image's own half-hourly cleanup deleting the compressed archives behind it. Audit logs now have a single owner, which keeps the live file and two rotations and compresses everything older
• Fix the console wizard's time server prompt opening empty whatever was configured, where confirming it then left a statically addressed host with no time source at all. This matches the reports of time synchronisation stopping after the move to chrony
• Fix a host whose name was set at first boot rather than through the wizard logging a resolution failure on every administrative command, as both a journal record and an audit event
• Fix the dashboard hiding host warnings and the new-version notice on a server with no products deployed yet, which is the server most likely to need them
• Fix a deploy reporting the wrong number of running containers, by counting the copies that are wanted and the ones still starting rather than every task the stack has ever had
• Fix the log viewer failing to start where its container could not write temporary files
• Fix the host upgrade log losing everything written from inside a container, so an upgrade that failed there left no record of why

Changed

• Saving a Portal or Core connection that does not answer is now a warning rather than a refused save, since the component being pointed at is routinely not deployed yet. The check re-runs in the background and the warning clears on its own once the connection works
• Audit records are kept on the host by default and are no longer copied into the system log as well, where the second copy could fill a partition sized for one

Security

• Harden the directories that containers mount from the host: they are root-owned, sticky and closed to other accounts, so a container that can write into one can no longer remove files it does not own. Each container's access is narrowed to what it needs, entries left behind by earlier releases are swept on every apply, and the hardening is re-applied at boot and after each deploy
• Audit records are retained for a week rather than four days, by excluding two kinds of daemon churn that accounted for almost all of the volume
• Update VM kernel 6.1.177 → 6.1.180 (cloud), delivering the latest Debian LTS kernel security fixes, including local use-after-free / privilege-escalation fixes (e.g. posix-cpu-timers exec race CVE-2026-64560 and cgroup writeback CVE-2026-64378) and a virtio-net receive-path fix (CVE-2026-64552). Requires a reboot.
• Upgrade the Installer container with the Debian 13.6 security batch — libpq5 (PostgreSQL client), util-linux / login, cryptography, pyasn1, soupsieve, sqlparse, liblzma5 and OpenSSL.
• Related CVEs: GHSA-537c-gmf6-5ccf, CVE-2026-6464, CVE-2026-6471, CVE-2026-6473, CVE-2026-14662, CVE-2026-14664, CVE-2026-14668, CVE-2026-14669, CVE-2026-14670, CVE-2026-14671, CVE-2026-14677, CVE-2026-14679, CVE-2026-14680, CVE-2026-15741, CVE-2026-15742, CVE-2026-16239, CVE-2026-18408, CVE-2026-19385, CVE-2026-49476, CVE-2026-49477, CVE-2026-53612, CVE-2026-53613, CVE-2026-53614, CVE-2026-54284, CVE-2026-59884, CVE-2026-59885, CVE-2026-59886, CVE-2026-59893, CVE-2026-63829, CVE-2026-64352, CVE-2026-64375, CVE-2026-64378, CVE-2026-64530, CVE-2026-64531, CVE-2026-64552, CVE-2026-64560, CVE-2026-69247, CVE-2026-69249, CVE-2026-71491

[1.8.20] - 2026-07-30

Fixed

• Fix docker container isolation issue that causes timeouts after 30 minutes when running Core and Portal on the same VM
• Fix grub bootloader EFI + Secure Boot issue during grub upgrade

Security

• Update VM kernel 6.1.176 → 6.1.177 (cloud), delivering the latest Debian LTS kernel security fixes. Includes a local privilege-escalation fix in the KEYS/keyctl subsystem (CVE-2026-63824). Requires a reboot.
• Upgrade grub2 boot loader (2.06-3~deb11u6 → ~deb11u7) — hardens the boot chain against crafted-filesystem parsing bugs (heap/integer overflows in the HFS/HFS+/UDF/squash4/JFS/BFS/UFS/tar/JPEG parsers and two lockdown bypasses). Boot-time/local integrity hardening, not network-reachable.
• Upgrade xz-utils / liblzma5 (decompression denial-of-service hardening)
• Related CVEs: CVE-2024-45782, CVE-2025-0624, CVE-2025-0678, CVE-2025-0689, CVE-2025-1125, CVE-2026-53157, CVE-2026-53366, CVE-2026-63795, CVE-2026-63824, CVE-2026-63830, CVE-2026-64191

[1.8.19] - 2026-07-12

Fixed

• Fix upgrade delay issues from some versions where you had to wait for the installer to finish or re-press upgrade multiple times

[1.8.18] - 2026-07-09

Added

• Add support to enable only FIPS-validated HTTPS crypto algorithms in load balancer
• Preparation for upcoming STIG hardening and FIPS support (separate license required) - contact us for information
• Add support for docker userns rewrite (see "Security below")

Fixed

• Force apply VM upgrades after updating Installer if the first run fails
• Ignore NXDOMAIN errors when resolving hostnames for health check
• Filter auditd rules before applying file monitoring to only include rules that are relevant to the current system
• Fix resetting installer password using cli component passwd without specifying username
• Fix installer password after database restore

Security

• Upgrade VM kernel from 5.10 to 6.1 (cloud), delivering the accumulated Debian kernel security fixes. This includes the full fix for CVE-2026-46300 ("Fragnesia"), for which 1.8.17 only shipped an interim mitigation. Requires a reboot.
• Upgrade openssl, libssl1.1, libgnutls30, krb5 (libgssapi-krb5-2, libk5crypto3, libkrb5-3, libkrb5support0), sudo and hyperv-daemons
• Related CVEs: CVE-2022-50552, CVE-2023-53596, CVE-2025-10263, CVE-2025-21863, CVE-2025-22107, CVE-2025-38129, CVE-2025-38584, CVE-2026-6473, CVE-2026-6477, CVE-2026-6478, CVE-2026-23099, CVE-2026-23372, CVE-2026-23397, CVE-2026-31399, CVE-2026-31452, CVE-2026-31455, CVE-2026-31485, CVE-2026-31489, CVE-2026-31580, CVE-2026-31607, CVE-2026-31627, CVE-2026-31657, CVE-2026-31659, CVE-2026-31686, CVE-2026-31696, CVE-2026-31720, CVE-2026-31747, CVE-2026-31748, CVE-2026-31759, CVE-2026-31787, CVE-2026-31788, CVE-2026-35535, CVE-2026-43015, CVE-2026-43027, CVE-2026-43040, CVE-2026-43114, CVE-2026-43196, CVE-2026-43281, CVE-2026-43328, CVE-2026-43386, CVE-2026-43426, CVE-2026-43427, CVE-2026-43449, CVE-2026-43450, CVE-2026-43453, CVE-2026-43458, CVE-2026-43503, CVE-2026-45866, CVE-2026-45867, CVE-2026-45879, CVE-2026-45885, CVE-2026-45914, CVE-2026-45916, CVE-2026-45920, CVE-2026-45936, CVE-2026-45984, CVE-2026-45994, CVE-2026-46022, CVE-2026-46033, CVE-2026-46047, CVE-2026-46064, CVE-2026-46191, CVE-2026-46243, CVE-2026-46285, CVE-2026-46294, CVE-2026-46300, CVE-2026-46301, CVE-2026-52914, CVE-2026-52943, CVE-2026-52972, CVE-2026-53002, CVE-2026-53130, CVE-2026-53266, CVE-2026-53294, CVE-2026-53296
• Use SCRAM- for authentication to internal PostgreSQL (database server is still not network accessible)
• Isolate docker containers further when running in shared networks (e.g. traefik load balancer ingress)
• Add support for docker userns rewrite for increased security hardening using cli docker enable-userns. (beta) Note: this will rewrite all existing data on disk. Make sure to have a full VM backup before trying this in an existing instance

Changed

• Decrease DNS timeout and retries for a better experience in environment with enabled DNS recursion but no available network paths

[1.8.17] - 2026-05-22

Fixes

• Fix changing static IP for second NIC using onboard wizard
• Display available IPs in select box in Server network settings

Security

• Update linux kernel
• Related CVEs: CVE-2026-46333, CVE-2026-46300
• Applies mitigation for CVE-2026-46300 ("Fragnesia") until a new kernel is released printf 'install esp4 /bin/false\ninstall esp6 /bin/false\ninstall rxrpc /bin/false\n' | sudo tee /etc/modprobe.d/dirtyfrag.conf if sudo lsmod | egrep -q 'esp4|esp6|rxrpc'; then sudo rmmod esp4 esp6 rxrpc ; echo 1 | sudo tee /proc/sys/vm/drop_caches; fi

[1.8.16] - 2026-05-09

Security

• Update linux kernel - includes fix for CVE-2026-43284, CVE-2026-43500 ("Dirty Frag")
• Related CVEs: CVE-2026-43284, CVE-2026-43500

[1.8.15] - 2026-05-01

Fixed

• Stop printing running status if deploying a mix of active components and "Uninstalled but keep data" ones

Security

• Update linux kernel - includes fix for CVE-2026-31431 (Copy Fail). This update requires a reboot.
• Related CVEs: CVE-2026-31431
• Mividas components do not permit arbitrary user code execution by default, but we nonetheless recommend upgrading as soon as possible to reduce the attack surface exposed by any as-yet-unknown vulnerabilities

[1.8.14] - 2026-04-30

Added

• Add support to reboot VM from web gui
• Do DNS resolution health check on web gui start screen
• Add view to display the largest database tables sizes for installed components from web gui
• Add support to set private key type when generating new certificates/CSRs
• Add CLI-command to clear redis-data for a container in case of corruption cli component redis_clear
• Improved handling for flushing write caches in data layer during hypervisor snapshot event

Fixed

• Fix cli db dump-command
• Fix serialization of encrypted SAML private keys
• Fix issue installing Core v4.1 on first generation VMs (~2019)

Security

• Add beta-support for changing container runtime to crun instead of runc to workaround CVE-2025-31133, CVE-2025-52565, CVE-2025-52881 (although these vulnerabilities are not really applicable for Mividas appliance with pre-screened images, only for generic container platforms during untrusted image startup)
• Upgrade bind9-host, bind9-libs, containerd, django, hyperv-daemons, libpng16-16, libxml, libssl1.1, libsystemd0, libudev1, linux-image, openssh-client, openssh-server, openssh-sftp-server, openssl, python3-urllib3, python3.9, systemd, systemd-sysv, systemd-timesyncd, udev
• Related CVEs: CVE-2022-37454, CVE-2024-25621, CVE-2025-13836, CVE-2025-22121, CVE-2025-38201, CVE-2025-39702, CVE-2025-57833, CVE-2025-64458, CVE-2025-64459, CVE-2025-68366, CVE-2025-68724, CVE-2025-69419, CVE-2025-69421, CVE-2025-71091, CVE-2025-71093, CVE-2025-71123, CVE-2025-8194, CVE-2026-1299, CVE-2026-1519, CVE-2026-2004, CVE-2026-2005, CVE-2026-2006, CVE-2026-21441, CVE-2026-22695, CVE-2026-22801, CVE-2026-22998, CVE-2026-23001, CVE-2026-23074, CVE-2026-23083, CVE-2026-23209, CVE-2026-23268, CVE-2026-23408, CVE-2026-23410, CVE-2026-23411, CVE-2026-23949, CVE-2026-24049, CVE-2026-25646, CVE-2026-26007, CVE-2026-27459, CVE-2026-29111, CVE-2026-30922, CVE-2026-33416, CVE-2026-33636, CVE-2026-3497, CVE-2026-41066, CVE-2026-6100

[1.8.13] - 2026-01-15

Added

• Add support for including latest stable Installer in offline bundle by adding ?include_installer=1 to browser URL before pressing "Export offline bundle"
• Add support to configure SNMPv3 authentication from wizard

Fixed

• Fix SNMP configuration in wizard
• Fix problem where offline installations stops on "missing offline bundle"-error. To upgrade Installer, import bundle and either reboot or run upgrade command from cli wizard
• Fix displaying EULA in offline mode

Security

• Update SSH cryptographic algorithms to modern standards

Changed

• Display warning indicator next to expired Intermediary CA certificates
• Change help message about maximum length for SMS sender option

[1.8.12] - 2025-12-22

Added

• Pass information about any trial licenses to each component

Fixed

• Match pre-released offline versions of the same component if license has access to multiple channels
• Verify connection for any Identify to Portal API-connections

Security

• Enable HSTS-headers for Installer port and, with short duration, any non-bound FQDNs
• Patch Django for CVE-2025-64460, CVE-2025-133722
• Upgrade libpng16-16, urllib3, cryptography, libssl
• Related CVEs: CVE-2025-64720, CVE-2025-65018, CVE-2025-66293, CVE-2025-66418, CVE-2025-66471

Changed

• Use custom logic for disk cleanup in web gui to prevent docker from removing images used by components
• Enable basic syntax highlighting for log output

[1.8.11] - 2025-10-21

Fixed

• Fix EC private key import
• Stop generating certificates for uninstalled components that are kept in list

Security

• Upgrade bind9-host, bind9-libs, hyperv-daemons, libfastjson4, libgnutls30, libpam-modules, libpam-modules-bin, libpam-runtime, libpam0g, libxml2, libxslt1.1, linux-image-5.10.0-35-cloud-amd64, linux-image-cloud-amd64, open-vm-tools, qemu-guest-agent, sudo
• Related CVEs: CVE-2025-40778, CVE-2025-40780, CVE-2025-8677, CVE-2022-21546, CVE-2023-52935, CVE-2024-26739, CVE-2025-37798, CVE-2025-37819, CVE-2025-37890, CVE-2025-37913, CVE-2025-37914, CVE-2025-37915, CVE-2025-37923, CVE-2025-37927, CVE-2025-38000, CVE-2025-38001, CVE-2025-38052, CVE-2025-38075, CVE-2025-38079, CVE-2025-38084, CVE-2025-38085, CVE-2025-38095, CVE-2025-38100, CVE-2025-38107, CVE-2025-38108, CVE-2025-38112, CVE-2025-38115, CVE-2025-38138, CVE-2025-38146, CVE-2025-38147, CVE-2025-38160, CVE-2025-38177, CVE-2025-38180, CVE-2025-38181, CVE-2025-38184, CVE-2025-38197, CVE-2025-38200, CVE-2025-38206, CVE-2025-38212, CVE-2025-38313, CVE-2025-38324, CVE-2025-38332, CVE-2025-38350, CVE-2025-38352, CVE-2025-38386, CVE-2025-38399, CVE-2025-38424, CVE-2025-38439, CVE-2025-38457, CVE-2025-38459, CVE-2025-38464, CVE-2025-38466, CVE-2025-38474, CVE-2025-38477, CVE-2025-38498, CVE-2025-38569, CVE-2025-38572, CVE-2025-38666, CVE-2025-38718, CVE-2025-39817, CVE-2025-39824, CVE-2025-39828, CVE-2025-39841, CVE-2020-12762, CVE-2025-32988, CVE-2025-32990, CVE-2025-6020, CVE-2025-49794, CVE-2025-49796, CVE-2025-6021, CVE-2025-7424, CVE-2025-41244, CVE-2024-7409, CVE-2025-32462

Changed

• Drop ICMP timestamp request packages

[1.8.10] - 2025-06-18

Added

• Try pinging host using different packet sizes in network tools to help find MTU-issues
• Add settings to enable SAML for backend admin against Mividas Identify v1.1.0+
• Add support to disable exporting private key of self signed certificates from UI.
• Add support to redact IP addresses from log files

Fixed

• Ping command in network tools
• Delay initial boot to give NTP some time to sync before generating certificates based on current time
• Validate certificate/private key combination when updating one at the time
• Skip importing already existing CAs during full chain imports
• Fix until-filter for docker service logs
• Fix false positive message about running a pre-release version of Installer on upgrade page
• Fix SAML group membership filter - it was previously not applied correctly.
• Fix using offline XML metadata for SAML
• Fix parsing log lines with timestamps without year if the date is from last year
• Fix redeploying all components using CLI
• Fix cleanup of audit logs for installations with lots of rotated files
• Fix SNMP configuration in wizard
• Fix changing password using cli component passwd
• Clear any corrupt docker gzipped log files to prevent parse errors during report
• Fix file permission on VM state file to allow status/warning messages to be seen in Installer on all installation
• Collect log from each docker service task individually to workaround docker bug which sometimes crashes dockerd when trying to pull logs from failed tasks
• Fix printing syslog when having HTTP proxy enabled

Security

• Fix SAML group membership filter - it was previously not applied correctly. Mitigation for older versions is to limit logins in the idP
• Upgrade Installer dependencies: gunicorn, libexpat1, libxml2
  • Related CVEs: CVE-2024-6827, CVE-2024-45491, CVE-2024-45492, CVE-2024-56171, CVE-2024-56171
• Upgrade VM libraries: bind9-libs, busybox, docker.io, hyperv-daemons, libcap2, libfreetype6, libglib2.0-0, libperl5.32, libxml2, libxslt1.1, linux-image-cloud-amd64, perl-base, perl-modules-5.32, python3.9, vim-common, vim-tiny, xxd
  • Related CVEs: CVE-2024-11187, CVE-2022-48174, CVE-2021-28831, CVE-2021-42378, CVE-2021-42379, CVE-2021-42380, CVE-2021-42381, CVE-2021-42382, CVE-2021-42384, CVE-2021-42385, CVE-2021-42386, CVE-2024-41110, CVE-2024-47685, CVE-2024-38538, CVE-2024-38588, CVE-2024-44940, CVE-2024-46853, CVE-2024-46854, CVE-2024-46858, CVE-2024-46865, CVE-2024-47701, CVE-2024-47742, CVE-2024-47748, CVE-2024-49860, CVE-2024-49882, CVE-2024-49883, CVE-2024-49884, CVE-2024-49889, CVE-2024-49983, CVE-2024-49995, CVE-2024-50033, CVE-2024-50036, CVE-2024-50055, CVE-2024-50083, CVE-2024-50115, CVE-2024-50127, CVE-2024-50131, CVE-2024-50193, CVE-2024-50262, CVE-2024-50278, CVE-2024-50279, CVE-2024-50301, CVE-2024-53057, CVE-2024-53096, CVE-2024-53099, CVE-2024-53103, CVE-2024-53141, CVE-2024-53142, CVE-2024-53227, CVE-2024-56601, CVE-2024-56606, CVE-2024-56615, CVE-2024-56631, CVE-2024-56642, CVE-2024-56650, CVE-2024-56662, CVE-2024-56672, CVE-2024-56766, CVE-2024-57951, CVE-2025-21648, CVE-2025-21687, CVE-2025-21692, CVE-2023-2603, CVE-2025-27363, CVE-2024-52533, CVE-2020-16156, CVE-2023-31484, CVE-2021-29921, CVE-2015-20107, CVE-2020-10735, CVE-2021-28861, CVE-2021-3737, CVE-2022-0391, CVE-2022-42919, CVE-2022-45061, CVE-2023-24329, CVE-2023-6597, CVE-2024-6232, CVE-2024-7592, CVE-2024-9287, CVE-2024-25062, CVE-2024-56171, CVE-2025-24928, CVE-2025-27113, CVE-2024-55549, CVE-2025-24855, CVE-2023-49083, CVE-2023-43804, CVE-2021-3872, CVE-2021-4019, CVE-2021-4173, CVE-2021-4187, CVE-2022-0261, CVE-2022-0351, CVE-2022-0359, CVE-2022-0361, CVE-2022-0392, CVE-2022-0417, CVE-2022-0572, CVE-2022-1616, CVE-2022-1785, CVE-2022-1897, CVE-2022-1942, CVE-2022-2000, CVE-2022-2129, CVE-2022-2304, CVE-2022-3099, CVE-2022-3134, CVE-2022-3324, CVE-2022-4141, CVE-2023-0054, CVE-2023-2610, CVE-2023-4738, CVE-2023-4752, CVE-2023-4781, CVE-2023-5344, CVE-2024-22667

Changed

• Upgrade cloud-init to 22.4.2 for new installations
• Sort service names in log view
• Link to certificate page to manage certificates instead of displaying inline file upload inputs in forms
• Reload certificate lists in select boxes automatically
• Use correct help text for Identify API-field, add more information about SAML and LDAP
• Display message about necessary reboot during uninstall if containerd is stuck
• Display logs full width in page, stream output in chunks for service status and log lines
• Use service task id for the instance of each log entry in log output

[1.8.9] - 2024-10-09

Added

• Add function to clean up old container images from the web gui server menu

Fixed

• Fix HTTP -> HTTPS redirect
• Fix building certificate chain during deploy when Intermediate but not Root CA is included in certificate file
• Store subject of Intermediate CA as name in the database, instead of Issuer, when importing CA-chain
• Always use service definition for the correct version when installing components in offline mode
• Fix conflicts when pinning image used in multiple components (e.g. postgres) in offline mode
• Abort offline installation in case of missing offline bundles
• Fix audit log max file number for new installation
• Fix upgrading specific Installer version from CLI
• Save snapshot queue reset-script in filesystem on upgraded installations
• Don't try to parse SAML claim names in URL-format as search+replace fields
• Use updated license flags as soon as possible, instead of after each deploy

Security

• Upgrade bind9-host, bind9-libs, e2fsprogs, hyperv-daemons, libcom-err2, libexpat1, libext2fs2, libgssapi-krb5-2, libk5crypto3, libkrb5-3, libkrb5support0, libsqlite3-0, libss2, libsystemd0, libudev1, libxml2, linux-image-5.10.0-30-cloud-amd64, linux-image-cloud-amd64, logsave, python3-pkg-resources, python3-setuptools, runc, shim-unsigned, systemd, systemd-sysv, systemd-timesyncd, traefik, udev
• Related CVEs: CVE-2021-36690, CVE-2022-1304, CVE-2022-2309, CVE-2022-43945, CVE-2022-48733, CVE-2023-27561, CVE-2023-28642, CVE-2023-40547, CVE-2023-40548, CVE-2023-50387, CVE-2023-50868, CVE-2023-52425, CVE-2023-7104, CVE-2024-1737, CVE-2024-1975, CVE-2024-27397, CVE-2024-36971, CVE-2024-36978, CVE-2024-37370, CVE-2024-37371, CVE-2024-38381, CVE-2024-38555, CVE-2024-38577, CVE-2024-38627, CVE-2024-39480, CVE-2024-39487, CVE-2024-4076, CVE-2024-40958, CVE-2024-41000, CVE-2024-41040, CVE-2024-41046, CVE-2024-41049, CVE-2024-41070, CVE-2024-41087, CVE-2024-41090, CVE-2024-41091, CVE-2024-42148, CVE-2024-42284, CVE-2024-42285, CVE-2024-42301, CVE-2024-42302, CVE-2024-42313, CVE-2024-43882, CVE-2024-44974, CVE-2024-44987, CVE-2024-44998, CVE-2024-44999, CVE-2024-45490, CVE-2024-45491, CVE-2024-45492, CVE-2024-46738, CVE-2024-46740, CVE-2024-46743, CVE-2024-46744, CVE-2024-46747, CVE-2024-46782, CVE-2024-46800, CVE-2024-46844, CVE-2024-6345

Changed

• Require confirmation when upgrading to another major version
• Stop writing cron events to audit logs
• Automatically reset expired Swarm-certificate for VMs that have been shutdown too long
• Pre-fill domain names in CSR-form
• Display timezone for log viewer fields
• Include digest information for each image in offline bundle-export
• Use CONNECT with HTTP/1.1 including Host-header when using HTTP(S) proxy

[1.8.8] - 2024-07-01

Fixed

• Increase timeout in load balancer (traefik) CVE mitigation that caused issue with large file uploads
• Truncate name of container secret if name of instance is too long

[1.8.7] - 2024-05-29

Added

• Add script to restore scheduled tasks after VM snapshot/backup-restore

Fixed

• Fix intermittent error message on dashboard about acquiring lock
• Fix SRV-lookup for LDAP servers
• Fix syntax of AD LDAP filter example in network tools
• Hide validation error alert for unfilled optional forms during installation
• Fix resetting fallback user password using sudo cli component passwd
• Fix file permissions in upgrade container
• Freeze auditd rules for existing installations
• Fix starting ssh directly on boot during initial installation

Security

• Upgrade bsdutils, django, dnspython, fdisk, gunicorn, less, libblkid1, libexpat1, libfdisk1, libmount1, libsmartcols1, libuuid1, linux-kernel, gunicorn, mount, paramiko, sqlparse, traefik, tzdata, util-linux
• Related CVEs: CVE-2021-37600, CVE-2023-29483, CVE-2022-48624, CVE-2024-32487, CVE-2023-48795, CVE-2023-52425, CVE-2024-1135, CVE-2024-24680, CVE-2024-28085, CVE-2024-4340, CVE-2024-2961, -1

Changed

• Use pool for database connections
• Upgrade load balancer (traefik) to v2.11
• Use postgresql 16 for new installations of both Installer and components
• Improve cache handling for SAML metadata. Add HTTP PRoxy-support
• Set upper process limit for containers
• Disable routing for internal load balancer network on new installations

[1.8.6] - 2024-03-22

Fixed

• Fix locking issue during metadata update that sometimes causes long timeouts and overuse of database connections after enough user actions without restart. Run sudo cli installer upgrade from SSH to upgrade if that happens, or restart the server
• Fix deploy of LDAPAdmin when using certificate
• Fix displaying certificate error message on screen
• Fix LDAP username attribute when using search+replace

Security

• Upgrade libuv1
• Related CVEs: CVE-2024-24806

[1.8.5] - 2024-03-05

Added

• Resolve LDAP servers in network tools test. Try to verify SSL-connection
• Add support for filtering log with until-filter
• Add support to disable SAML-login for installer using "cli installer disable_saml" and "cli installer restore_saml"

Fixed

• Fix fetching syslog in Logs-view
• Don't enable SAML SLO if upgrading from 1.8.3 or earlier
• Use more conservative upscaling of services to decrease risk of Out of memory-errors
• Increase global max memory limit to prevent VM scaling to increase potential memory usage too much in case of using 12+ GB of RAM
• Remove external version check-request from internal load balancer
• Fix unnecessary restarts of Installer-component

Security

• Upgrade hyperv-daemons, libglib2.0-0, libgnutls30, libperl5.32, linux-image-5.10.0-26-cloud-amd64, linux-image-cloud-amd64, perl-base, perl-modules-5.32, runc
• Related CVEs: CVE-2023-29499, CVE-2023-35827, CVE-2023-46813, CVE-2023-47038, CVE-2023-5178, CVE-2023-5717, CVE-2023-6040, CVE-2023-6531, CVE-2023-6817, CVE-2023-6931, CVE-2023-6932, CVE-2024-0567, CVE-2024-0646, CVE-2024-1086, CVE-2024-2162,

Changed

• Set flag in SAML SP-metadata about wanting assertions signed
• Use streaming console output for long-lasting commands
• Change console log-format
• Refresh session expiry on each request
• Include extra information about running services in log output
• Use logrotate for auditd-logs
• Limit redundant journald-log size

[1.8.4] - 2023-12-31

Added

• Add support for SAML login. See /saml/metadata/ for SP metadata.
• Display EULA in interface
• Add support to configure remote syslog server from onboarding wizard
• Add support to enable SNMP monitoring (new installations only for now)
• Add support to configure remote syslog server from onboarding wizard
• Add support to enable SNMP monitoring (new installations only for now)

Fixed

• Fix syntax in example AD LDAP-filter
• Fix permissions of uploads-folder if containers was started in wrong order
• Remove log files if log partition is full
• Fix permissions of uploads-folder if containers was started in wrong order
• Remove log files if log partition is full
• Fix sorting of syslog log lines

Security

• Upgrade django, libnghttp2-14, openssh-client, openssh-server, openssh-sftp-server, libpq5
• Related CVEs: CVE-2023-44487, CVE-2023-51385, CVE-2021-41617, CVE-2023-46695, CVE-2023-5869

Changed

• Use license key from uploaded file if both textarea and file is provided at the same time
• Automatically remove newline and spaces from license key before validation
• Use more memory for each worker when calculating max number of workers for VM-size
• Limit memory to max 50% of total VM memory for each service
• Change OOM score for database-containers
• Increase swap space for older VM-versions
• Change OOM score for database-containers
• Increase swap space for older VM-versions

[1.8.3] - [2023-10-31]

Added

• Add support to use smtp+tls for SMTP relay
• Add separate field for required LDAP-group for normal users
• Add support to set SAML claims to use for user/admin/superuser permission
• Add support to set SAML username claim, with support for search+replace
• Add support to force restart of services in log view
• Display disk space warnings on dashboard
• Display available Installer/VM-upgrades on dashboard
• Display if reboot is necessary on dashboard
• Allow specifying specific Installer version when upgrading from command line
• Add support to display syslog in web-UI

Fixed

• Fix updating component reference if converting between licensed products
• Keep currently installed component version in version select-box if it has been removed from global list
• Allow disabling SAML signing certificate
• Don't display certificates belonging to non-deployed components as active
• Fix update of overwritten dev-versions during deploy in offline mode
• Fix quoting of special characters in HTTP(S) Proxy authentication
• Fix system logs rotation-timer
• Upgrade libseccomp for Debian buster based installations to allow for bookwork based docker-images
• Fix setting static routes on second NIC if routes is configured later

Security

Upgrade bind9-host, bind9-libs, cpio, curl, hyperv-daemons, libc-bin, libc-l10n, libc6, libcurl4, libjson-c5, libncurses6, libncursesw6, libtinfo6, linux-image-5.10.0-23-cloud-amd64, linux-image-cloud-amd64, locales, ncurses-base, ncurses-bin, open-vm-tools, openssh-client, openssh-server, openssh-sftp-server, qemu-guest-agent

• Related CVEs: CVE-2021-32292, CVE-2021-38185, CVE-2022-39189, CVE-2023-1989, CVE-2023-20900, CVE-2023-2156, CVE-2023-29491, CVE-2023-3090, CVE-2023-31248, CVE-2023-3268, CVE-2023-3341, CVE-2023-3354, CVE-2023-3389, CVE-2023-3390, CVE-2023-35001, CVE-2023-35788, CVE-2023-3609, CVE-2023-3610, CVE-2023-3611, CVE-2023-3776, CVE-2023-3777, CVE-2023-38408, CVE-2023-4004, CVE-2023-40283, CVE-2023-4128, CVE-2023-4147, CVE-2023-4206, CVE-2023-4207, CVE-2023-4208, CVE-2023-4244, CVE-2023-42753, CVE-2023-4622, CVE-2023-4623, CVE-2023-4911, CVE-2023-4921, CVE-2023-38325, CVE-2023-37920

Changed

• Automatically retry installation if dockerd times out
• Hide gunicorn version from Server-header
• Display additional disclaimer about LDAP user access
• Revert to old postgres version by default
• Auto-select only services that are running too many or too few instances in Log view
• Split CA certificates to multiple rows during import. Don't import duplicates
• Try to add missing CA-certificates to certificate chain if missing from component certificate file
• Remove name-field from CA import form. Automatically use subject
• Compress logs the first rotation
• Change default LDAP filter to not include disabled users when using AD

[1.8.2] - 2023-07-12

Added

• Add new TLS cipher option with modern ciphers plus two weak TLS 1.2-ciphers to allow for older firmware of external systems
• Add support for separate component containing only Mividas Scheduling Portal
• Allow converting between the same product families e.g. for testing development releases
• Add support for providing SAML certificate/key pair
• Add support for selecting TLS-mode and authentication method for external databases

Fixed

• Don't try to use broken/invalid server default certificate/private key-pairs
• Use default server timezone when parsing log time filter
• Fix help text of certificate textbox since adding support for importing encrypted private keys
• Run log rotation more often for container logs
• Discard some non-relevant kernel logs during container initialization and teardown
• Escape http proxy username/password in environment variables
• Use container service name in syslog
• Fix database dump using "cli"-command

Security

• Upgrade curl, linux kernel, openssl, containerd, sudo
• Upgrade redis, requests, setuptools, sentry-sdk
• Related CVEs: CVE-2023-25173, CVE-2022-32221, CVE-2023-0179, CVE-2023-1077, CVE-2023-1281, CVE-2023-1829, CVE-2023-1872, CVE-2023-32233, CVE-2023-0361, CVE-2022-29458, CVE-2022-4450, CVE-2023-0215, CVE-2023-0286, CVE-2023-0464, CVE-2023-2650, CVE-2023-22809,
• Disable network routing for internal container services
• Set session expiration to 2h
• Fix issue in error reporting where the system tried to send a report of number of discarded error-messages to a remote server, due to that error reporting was disabled

Changed

• Change "Mividas Connect" to new product name "Mividas Identify"
• Remove TLS 1.0 from the most insecure TLS-cipher option
• Remove LDAP default server settings
• Pass host VM ca-certificates to containers
• Use postgres 13 for new installations
• Move swap file to the same partition as user data
• Increase default disk size to 200GB, increase log partition-size to 2.5GB
• Disable unattended installation of subset of debian security upgrades by default

[1.8.1] - 2023-01-15

Added

• Prepare SAML authentication settings for Mividas Core + Rooms

Fixed

• Fix home link from product details view
• Lock postgres version for LDAPAdmin deployed with old version of deploy file
• Stop re-deploying load balancer on installer upgrade if not necessary
• Remove warning about missing volumes when removing component
• Fix service deploy problems when using setting values (e.g. passwords) starting with quotes (")
• Fix host security update files when upgrading system using offline bundle

Changed

• Change deploy mode for some shared services to allow Installer upgrades in the future with less downtime
• Increase number of workers for each component based on available memory
• Always try to start Installer based on script from the currently running version when using other than the official latest version
• Encrypt API keys and passwords before saving to database
• Remove letsencrypt option from certificate config screen
• Limit access for logs and system files, fix some CIS benchmark warnings, enable console timeout
• Install host security upgrades just after upgrading Installer, stop docker from potentially being upgraded automatically
• Discard some recurring kernel log messages about virtual container network interfaces
• Allow more userdata in cloud-init config

Security

• Upgrade host packages for grub2, hyperv-daemons, libexpat1, libtasn1-6, libxml2, linux-image-cloud-amd64
• Related CVEs: CVE-2021-46848, CVE-2022-2601, CVE-2022-3524, CVE-2022-3565, CVE-2022-3594, CVE-2022-3775, CVE-2022-40303, CVE-2022-40304, CVE-2022-4139, CVE-2022-42896, CVE-2022-43680, CVE-2022-4378, CVE-2022-47518, CVE-2022-47519
• Limit system permissions for load balancer container, run more services with read only root file system

[1.8.0] - 2022-10-20

Added

• Add support to enable external database from license file
• Add support to importing and converting binary certificates (beta)
• Add support to export private keys with encryption, and importing encrypted keys
• Add support for external redis server
• Add support for enabling LDAP referral chasing
• Add support to lookup LDAP servers using SRV records
• Add more inline documentation for LDAP, SAML and SMS-settings
• Mark certificates in use in lists
• Add field for validating SSL handshake against remote port using network tools
• Validation of database and ldap settings when saving configuration
• Display information about last component metadata refresh time and add link to force refresh
• Add CLI command ("cli") with support to, among other things, reset passwords and dump database content
• Allow ICMP echo requests ("ping")
• Allow overriding dns when using DHCP
• Install traceroute
• Add support for EFI and Secure Boot (beta, new VM installations only)

Fixed

• Clear certificate existing CA chain when updating public key
• Fix TLS validation for LDAP test connection
• Validate line before removing volumes
• Fix version ordering for x.y.z-dev builds
• Fix permission to run ping in network tools
• Fix certificate chain warnings if CA lacks common name information
• Reset offline mode if online license validation was successful
• Increase length of ldap filter
• Remove warning in load balancer logs about SNI host
• Stop validating values in optional forms marked for deletion
• Fix offline export if "check for update"-checkbox is not set
• Fix offline export if any components are marked for uninstallation
• Remove console log warnings about HostSNI
• Set static routes after all interfaces are up
• Install systemd-timesyncd if the initial VM version did not include it
• Fix returning to menu after setting hostname

Security

• Upgrade django, openssl, libssl1.1, sqlite3
• Related CVEs: CVE-2022-28346, CVE-2022-28347, CVE-2022-0778
• Upgrade bind9-host, bind9-libs, curl, dpkg, grub-common, grub-pc, hyperv-daemons, linux-image-cloud-amd64, libc-bin, libtasn1-6, libssl1.1, openssl, qemu-guest-agent, rsyslog, zlib1g
• Related CVEs: CVE-2018-13405, CVE-2021-22945, CVE-2021-22946, CVE-2021-30560, CVE-2021-3697, CVE-2021-3999, CVE-2021-4197, CVE-2021-4206, CVE-2021-4207, CVE-2021-46828, CVE-2022-0358, CVE-2022-1012, CVE-2022-1158, CVE-2022-1292, CVE-2022-1353, CVE-2022-1586, CVE-2022-1587, CVE-2022-1652, CVE-2022-1664, CVE-2022-1679, CVE-2022-1729, CVE-2022-1786, CVE-2022-20368, CVE-2022-20422, CVE-2022-20566, CVE-2022-20568, CVE-2022-2068, CVE-2022-22576, CVE-2022-2327, CVE-2022-24903, CVE-2022-2509, CVE-2022-2585, CVE-2022-2588, CVE-2022-2602, CVE-2022-26353, CVE-2022-27404, CVE-2022-27405, CVE-2022-27406, CVE-2022-27666, CVE-2022-27775, CVE-2022-27781, CVE-2022-27782, CVE-2022-2795, CVE-2022-28733, CVE-2022-28734, CVE-2022-29155, CVE-2022-29162, CVE-2022-29581, CVE-2022-29582, CVE-2022-2959, CVE-2022-2977, CVE-2022-30594, CVE-2022-3080, CVE-2022-31676, CVE-2022-3176, CVE-2022-32207, CVE-2022-32250, CVE-2022-34918, CVE-2022-3625, CVE-2022-3635, CVE-2022-36946, CVE-2022-38177, CVE-2022-38178, CVE-2022-40674, CVE-2022-41222, CVE-2022-43750

Changed

• Only display first certificate chain warning, hide warning if three or more certificates are included
• Increase log verbosity for ldap tests
• Set ldap connection timeout
• List any unknown/not fully uninstalled container services
• Add pagination and search to certificate lists
• Use direct API for fetching service logs instead of subprocesses for better performance
• Escape special characters in authentication to external services
• Rotate log files more often
• Increase log file partition size (for new VMs)
• Decrease console log verbosity
• Change docker internal IP series to 100.64.10[3-5].0/16 to limit risk of conflicts (new VMs only)
• Change to GPT based partitions
• Add docker/-prefix to syslog tag, write container logs to separate files in /var/log/docker/
• Prepare for support for external syslog servers. Manual configuration should be moved to /etc/rsyslog.d/50-remote.conf
• Enable SSH login by default for new installation, enable fail2ban to lock logins after too many logins