Did it really run?

[bondskin]

Describe the bug

Hi,

Successfully installed the latest version yesterday.
Usually the job runs every night, never an issue before.
But today it did not run with the latest release.

Am I missing something obvious?

Steps to reproduce

Latest image pulled
Schedule Cron job

Expected behavior

Scan job writes a log entry

Screenshots

No response

Environment

• Debian
• amd64
• Docker Version

Your logs

time="2025-12-10T03:07:23+01:00" level=info msg="Detected multiple Watchtower instances, initiating cleanup" count=2

time="2025-12-10T03:07:23+01:00" level=info msg="Stopping container" container=CNZDEjSUGQvzeNxYBuIyByCSoMjAACcI id=8717e958a923 signal=SIGTERM timeout=10m0s

time="2025-12-10T03:07:25+01:00" level=info msg="Removing image" image_id=f42281e507ec image_name="nickfedor/watchtower:latest" notify=yes

time="2025-12-10T03:07:25+01:00" level=info msg="Successfully cleaned up excess Watchtower instances"

time="2025-12-10T03:07:25+01:00" level=info msg="Watchtower 1.12.4 using Docker API v1.51"

time="2025-12-10T03:07:25+01:00" level=info msg="Using notifications: smtp"

time="2025-12-10T03:07:25+01:00" level=info msg="Next scheduled run: 2025-12-11 03:05:00 CET in 23 hours, 57 minutes, 34 seconds"

Additional context

No response

[nicholas-fedor]

The logs that you shared show a newly updated Watchtower 1.12.4 container that is performing a normal self-cleanup after a self-update.

Generally speaking, a Watchtower container self-update occurs at the end of an update cycle, with the old/stale container creating the new one, changing the old container's name to a random string, and then finishing up the update process (i.e. sending notifications and etc).

Given the smtp notification setting, I would expect you to see a notification around the same time (2025-12-10T03:07:23+01:00) indicating corresponding update activity.

The expected "scan job log entry" would be expected in the old (probably removed) Watchtower container.

[bondskin]

Hi @nicholas-fedor ,

Thank you for the swift reply.
The self-update and such, this is all good.

I was explicitly worried about the missing scan job, or the absence of a log entry.
So this makes me difficult to triage if watchtower is running or not.

[nicholas-fedor]

Your concern is why many people have opted to utilize additional services, such as Portainer, Dozzle, etc to help with container and log management. This is a good jump-off point for understanding logging in Docker: https://docs.docker.com/engine/logging/configure/

When it comes to determining if Watchtower is running or not, there is a healthcheck that can be queried via the following command:

docker inspect --format='{{.State.Health.Status}}' <container_name_or_id>

Expected Output:

healthy

Then, there's always the option of waiting for the next scheduled run to see if the update process has run.

---

I'm moving this thread over to the discussions for the time being.

[bondskin]

Hi,

This is your work and I accept your decisions; I don't want to carry this further but I am still confused about the recent behaviour of the latest version.

Yesterday (Dec 12) it updated itself to latest version:

Watchtower 1.12.5 using Docker API v1.51
Using notifications: smtp
Next scheduled run: 2025-12-13 03:05:00 CET in 23 hours, 58 minutes, 11 seconds

Tonight (Dec 13) it did nothing, nothing in the logs, no containers were updated.
I re-created the container and then I got this long list of updated images

Found new image: ghcr.io/home-assistant/home-assistant:latest (6c287fba612d)
Found new image: lscr.io/linuxserver/heimdall:latest (7b3bbc1afd8d)
Found new image: lscr.io/linuxserver/swag:latest (7f998c95c40f)
Found new image: lscr.io/linuxserver/jackett:latest (40588b180013)
Found new image: ghcr.io/homarr-labs/homarr:latest (7b264978b347)
Found new image: homebridge/homebridge:latest (5dbecc641fd6)
Found new image: ghcr.io/paperless-ngx/paperless-ngx:latest (38cd7883d04f)
Found new image: amir20/dozzle:latest (b3f836eca3fd)

All updates went well, as always and as expected.
I just don't know why this did not work automatically tonight (this was working in the past without issues).

[nicholas-fedor]

Your feedback has been helpful, as it helps me to further refine (and sometimes break) things.
The behavior that you observed was due to logic that I introduced, which intentionally skips the new immediate run after a self-update. This was intended to specific to the use of the run-after-start option, but appears to impact broader update scheduling, such as preventing an update that's supposed to take place the following day. Needless to say, this will be improved.

[bondskin]

Thank you for your comprehensive feedback, much appreciated.
Last night the update did not occur as expected:

time=2025-12-17T03:06:48+01:00
level=info
msg=Watchtower 1.13.0 using Docker API v1.51
time=2025-12-17T03:06:48+01:00
level=info
msg=Using notifications: smtp
time=2025-12-17T03:06:48+01:00
level=info
msg=Next scheduled run: 2025-12-18 03:05:00 CET in 23 hours, 58 minutes, 11 seconds

I pulled the image and recreated the watchtower container this morning:

time=2025-12-18T10:27:12+01:00
level=info
msg=Watchtower 1.13.0 using Docker API v1.51
time=2025-12-18T10:27:12+01:00
level=info
msg=Using notifications: smtp
time=2025-12-18T10:27:12+01:00
level=info
msg=Update on startup enabled: Performing immediate check
time=2025-12-18T10:27:12+01:00
level=info
msg=Next scheduled run: 2025-12-19 03:05:00 CET in 16 hours, 37 minutes, 47 seconds
time=2025-12-18T10:27:23+01:00
level=info
msg=Found new image
container=traefik
image=traefik:latest
new_id=0fb158a64eaa
time=2025-12-18T10:27:28+01:00
level=info
msg=Found new image
container=jackett
image=lscr.io/linuxserver/jackett:latest
new_id=9877331b7a0b
time=2025-12-18T10:27:38+01:00
level=info
msg=Found new image
container=portainer
image=portainer/portainer-ce:latest
new_id=2622931a6f42

I enabled now the "--debug" command to continue observing the situation.