Update CHANGELOG for Mender Client 6.0.x

[mender-test-bot]

Mender Client 6.0.x

Repository	Version
mender	5.1.x
mender-connect	3.0.x
monitor-client	1.5.x
mender-flash	1.1.x
mender-configure-module	1.1.x
mender-binary-delta	1.5.x
mender-container-modules	1.0.x

mender 5.1.x (2026-08-21)

5.1.x - 2026-08-21

Bug fixes

• (auth) Fix segfault when dbus becomes unavailable and available again
  (MEN-9246) (7243c59)

  Change mender::api::HTTPClient::AsyncCall to void and handle all error callbacks asynchronously.
  Previously, errors in WithToken from StartWatchingTokenSignal were synchronous, and errors from
  GetJwtToken were asynchronous, but still returned an error. This caused the caller to not know
  wheter a callback would be called by AsyncCall, ultimately leading to the callback being called
  multiple times.
  Having AsyncCall not return anything, and always call the callback on error makes the
  contract cleaner. Each caller now knows that if there's an error in AsyncCall, whether
  it is immediate, or happens later after reading some data, the callback will be run
  exactly once.

• (http) Time out stalled transactions without killing switched protocols
  (MEN-9433MEN-10031) (d417a2a)

  A server that accepts a request and never answers left the response read
  outstanding forever, wedging the client until restart. Arm Beast's stream timeout
  per operation and disarm it at the 101 Switching Protocols handover -- leaving it
  armed there closed mender-connect's WebSocket after five minutes and got the
  previous attempt reverted (MEN-9433).

• (modules-artifact-gen) Fail early when mender-artifact produces an invalid artifact
  (80e7aa6)

• (modules-artifact-gen) Always remove temporary files
  (be174ef)

• (update-modules) Fail the install when the backup fails
  (MEN-10021) (9c64ef1)

  ret=$? was read inside if ! <backup>; then, where $? holds the status
  of the negated pipeline and is therefore always 0. Both modules exited 0
  before applying the payload, so a failed install was reported to the server
  as successful: the destination was left untouched while the new software
  version was recorded for it. Capture the status from the command itself
  instead.

  The directory module is the more exposed of the two, since it backs up the
  whole destination directory, so any unrelated file growing there can exhaust
  the data partition. Reported from the field with a multi-GB core dump.

  Add a pytest suite for the shipped update module scripts, alongside the
  existing one for the artifact generators, covering both a failing backup and
  a successful install for each module. Ignore the Python test caches while
  here, since the new suite is the second pytest directory in the tree.

• Don't retry status update on 413
  (ME-616) (e13437d)

• Handle first timestamp in deployment logs more carefully
  (MEN-9427) (68ab41e)

  Depending on the system and build configuration, the timestamps
  in logs can use lower (likely) or higher (unlikely) time
  resolution than expected (nanoseconds). So in case of a
  deployment failure with corrupted logs, when using the first
  timestamp as a replacement in the extra (THE ORIGINAL LOGS CONTAINED INVALID ENTRIES) log entry, care must be taken to make
  sure the extra log entry still has a valid timestamp and that the
  result is valid JSON.

• Sort inventory generator scripts before running them
  (MEN-9635) (a5fcec1)

  So that the results are consistent and don't depend on directory
  iteration ordering.

• Fail when device tier is invalid
  (ME-636) (590282f)

  DeviceTier is one of the configuration options but it is critical
  for how device is recognized by the server. When we will fail
  parsing it out of the configuration we should fail instead to
  switching to the standard device tier.

• Ensure limited permissions of mender-inventory-geo cache files
  (MEN-9752) (64584e2)

  The /tmp/mender/inventory-geo cache file can potentially contain
  sensitive data and there's no reason for it to have 644
  permissions.

  By using umask 077 we can easily ensure that any files created
  by the script are only user/owner-accessible.

• Default RetryPollIntervalSeconds to 300 seconds
  (MEN-9719) (9fc39d7)

  RetryPollIntervalSeconds defaulted to 0 seconds, meaning any config not
  setting RetryPollIntervalSeconds would retry polling with a 0 second
  delay. Default to 300 seconds, matching the Go client's fallback for unset
  values.

• Ensure corrent order of globals initalization
  (MEN-9878) (ca57f7b)

• Report download failure when DownloadResumerClient gives up mid-stream
  (MEN-9954) (19af738)

  When there are network issues during Artifact download and the
  HTTP download resumer exhausts its retry backoff, the only way to
  deliver the error to the state machine is through the HTTP body
  reader's handler because the body handler passed to the
  DownloadResumerClient is only called when all body data is
  fetched which normally happens in a completely different state
  than the one that starts the download and thus the
  UpdateDownloadState doesn't handle any errors in it.

  Add a Fail() function to the body reader and use it to make sure
  it calls its handler with the respective error.

  Also add a comment about how the DownloadResumerClient is used.

• Don't re-sanitize logs during upload
  (MEN-10017) (5465efd)

  Calling SanitizeLogs() as part of Rewind would invalidate the previous
  Content-Length calculation if the log has grown in the meantime. This
  would cause the server to reject the log upload with HTTP 400:
  unexpected EOF.

  The log file can grow between computing the Content-Length and streaming
  the body, because the deployment keeps logging to it while the upload is
  in flight.

Documentation

• Update contributing guide
  (QA-1517) (9372fc1)

  To remove custom extensions of conventional commits standard.

Features

• Large deployment logs are now trimmed to be accepted by the server
  (MEN-9415) (d870064)

  Deployment logs larger than 1 MiB are rejected by the
  server which leads to two issues:

  • excessive bandwith consumption when uploading such large logs
    only to be thrown away, and

  • no deployment logs for particular device and particular failed
    deployment available at the server at all.

  To prevent this, the client now trims large deployment logs and
  only sends the biggest possible part of the logs from their end.

mender-connect 3.0.x (2026-08-18)

3.0.x - 2026-08-18

Bug fixes

• Do not block the message loop when a late pong arrives
  (58d6a76)

  The healthcheck goroutine stops reading MenderShellSession.pong the
  moment it enters the timeout branch of its select, but the session stays
  reachable through MenderShellSessionGetById until MenderShellStopById
  removes it. Removal happens no earlier than four seconds later because
  procps.TerminateAndWait sleeps 2s + 2s unconditionally. A pong that
  arrives in that window finds the session, and HealthcheckPong blocks
  forever on an unbuffered channel that no longer has a reader.

  HealthcheckPong is called synchronously from routeMessagePongShell,
  which runs on the daemon's messageLoop goroutine. The block therefore
  stops the whole daemon: it stops reading the websocket, stops logging
  and never reconnects. The process stays alive, so systemd does not
  restart it.

  MenderShellStopById also returns without removing the session when
  StopShell fails and the shell process is still alive, which leaves the
  session reachable indefinitely and makes the same block reachable
  without any race.

  Give the channel a single slot and make the send non-blocking. A pong
  only signals that the peer is alive, so holding one is enough and
  dropping a second one changes nothing. The buffer keeps a pong that
  arrives while the healthcheck is momentarily outside its select, and the
  default case keeps the sender from blocking once the reader is gone.

  Add a regression test that delivers a pong inside the removal window,
  plus a negative control showing that a pong delivered after the removal
  is rejected with ErrSessionNotFound.

monitor-client 1.5.x (2026-07-31)

No changelog entries found.

mender-flash 1.1.x (2026-02-27)

No changelog entries found.

mender-configure-module 1.1.x (2026-05-28)

No changelog entries found.

mender-binary-delta 1.5.x (2026-08-06)

1.5.x - 2026-08-06

Build

• Update cloning link for cmocka, whose original has broken.
  (7cb08af)

mender-container-modules 1.0.x (2026-05-05)

1.0.x - 2026-05-05

Bug fixes

• Cd into manifest dir beforing listing running containers
  (MEN-9641) (0aa83f8)

  Fix an issue where docker-compose v1 can't find manifest files for a
  composition because we don't cd into the manifest directory.

[mender-test-bot]

@mender-test-bot, start a full client pipeline with:

• mentioning me and start client pipeline

---

my commands and options

You can prevent me from automatically starting CI pipelines:

• if your pull request title starts with "[NoCI] ..."

You can trigger a client pipeline on multiple prs with:

• mentioning me and start client pipeline --pr mender/127 --pr mender-connect/255

You can trigger a client pipeline for a specific Mender Client release with:

• mentioning me and start client pipeline --release 6.0.x (can be given multiple times)
• by default, a pipeline is triggered for each supported release the component is a part of

You can trigger GitHub->GitLab branch sync with:

• mentioning me and sync

You can print PR statistics for a repository with:

• mentioning me and print fast pr stats (Team stats only)
• mentioning me and print full pr stats (Detailed report)
• options: --repo <repo>, --team <name>, --all-repos, --exclude-drafts, --exclude-user <user>
• mentioning me and print full pr stats --repo mender --all-repos --exclude-drafts

You can cherry pick to a given branch or branches with:

• mentioning me and:

 cherry-pick to:
 * 1.0.x
 * 2.0.x

[mender-test-bot]

There was an error syncing branches, see logs for details.