Podman only: Switching between containers in the UI causes the container to crash.

[karami4yasser]

🔍 Check for existing issues

• Completed

How is Dozzle deployed?

Standalone Deployment

📦 Dozzle version

v8.13.14

✅ Command used to run Dozzle

panic: runtime error: slice bounds out of range [32:31]

goroutine 260 [running]:
github.com/amir20/dozzle/internal/docker.(*LogReader).Read(0xc000130e70)
        /dozzle/internal/docker/log_reader.go:62 +0xfd
github.com/amir20/dozzle/internal/container.(*EventGenerator).consumeReader(0xc0004cfaa0)
        /dozzle/internal/container/event_generator.go:85 +0x32
created by github.com/amir20/dozzle/internal/container.NewEventGenerator in goroutine 236
        /dozzle/internal/container/event_generator.go:46 +0x185

🐛 Describe the bug / provide steps to reproduce it

Switching between containers in the UI causes dozzle container to crash.

💻 Environment

oracle linux 8
podman version 4.9.4-rhel

📜 Dozzle logs are required for debugging purposes. You may need to enable debug mode. See https://dozzle.dev/guide/debugging.

{"level":"info","version":"v8.13.14","time":"2025-09-15T12:23:02Z","message":"Dozzle version v8.13.14"}
{"level":"info","version":"v8.13.14","clients":1,"time":"2025-09-15T12:23:03Z","message":"Connected to Docker"}
{"level":"info","version":"v8.13.14","time":"2025-09-15T12:23:03Z","message":"Accepting connections on :8080"}
panic: runtime error: slice bounds out of range [32:31]

goroutine 260 [running]:
github.com/amir20/dozzle/internal/docker.(*LogReader).Read(0xc000130e70)
        /dozzle/internal/docker/log_reader.go:62 +0xfd
github.com/amir20/dozzle/internal/container.(*EventGenerator).consumeReader(0xc0004cfaa0)
        /dozzle/internal/container/event_generator.go:85 +0x32
created by github.com/amir20/dozzle/internal/container.NewEventGenerator in goroutine 236
        /dozzle/internal/container/event_generator.go:46 +0x185

📸 If applicable, add screenshots to help explain your bug

[amir20]

I don't think this is related to switching between containers. I haven't seen any body else report this. Most likely it is related to something Podman is doing. The code:

message += tail[32:]  // I skip the first 32 chars because it is timestamp 

I am not sure how to fix this because I haven't actually seen it.

1. Are you sure it is when you switch containers? If you go directly to the container does it still do that?
2. Can you do export of the logs of the containers that is breaking this?

I could take a guess and try to fix it but I think something else is happening.

[karami4yasser]

Yes, this issue only happens when I start switching between containers (it usually fails on the second or third switch).

It starts working again only after restarting. Once it does, reaching dozzle/container/<id> does not cause it to crash.

Exporting logs also doesn’t seem to trigger the crash.

Please note that I have tested with & without :
DOZZLE_ENABLE_SHELL=true

[amir20]

Yes, this issue only happens when I start switching between containers (it usually fails on the second or third switch).

Hmm that is really weird.

Sorry I wasn't clear. Can you attach the logs? I can run it locally and see if something is happening.

🤔 I don't know what is causing this to be honest.

[amir20]

Also, not sure if you can help out, but test it on Docker? I am trying to narrow down if this is only happening on Podman or if it's something different with your cotnainers.

[ivoshm]

I've seen a similar bug in a similar environment (Podman), but unfortunately, I couldn't recreate the issue. It's probably related to the content of the log.

It would definitely help to verify that there are at least 32 characters in the tail variable before skipping it. Then, we would not get a panic, and we could see in Dozzle what Podman is actually sending.

[amir20]

Thanks @ivoshm . That helps to narrow. I'll have to setup podman on a linux vm and test it out.

It would definitely help to verify that there are at least 32 characters in the tail variable before skipping it. Then, we would not get a panic, and we could see in Dozzle what Podman is actually sending.

It's a little more complicated than that. Because that code should only happen for extremely large logs. That's why I need to reproduce to understand what is going on.

[amir20]

@ivoshm @karami4yasser I created a new Ubuntu VM and installed podman version 5.4.1. I did

podman run --volume /run/user/501/podman/podman.sock:/var/run/docker.sock -d -p 8080:8080 docker.io/amir20/dozzle:latest

Then did a bunch of fake containers. I switched around for 20+ times and nothing broke.

I would need to see the content of logs to help or somehow reproduce it.

[msaitz]

I started using dozzle with podman and I am encountering the same issue. The reproduction just took a few seconds while selecting different containers.

logs:

$ docker ps -a | grep dozzle
2f1fc37f22d8  docker.io/amir20/dozzle:latest                                      About a minute ago  Exited (2) 49 seconds ago  0.0.0.0:8089->8080/tcp                                                      dozzle

wyse:~$ docker logs dozzle
{"level":"info","version":"v8.14.12","time":"2025-12-27T22:46:21Z","message":"Dozzle version v8.14.12"}
{"level":"info","version":"v8.14.12","clients":1,"time":"2025-12-27T22:46:21Z","message":"Connected to Docker"}
{"level":"info","version":"v8.14.12","time":"2025-12-27T22:46:21Z","message":"Accepting connections on :8080"}
panic: runtime error: slice bounds out of range [32:27]

goroutine 273 [running]:
github.com/amir20/dozzle/internal/docker.(*LogReader).Read(0xc0000109c0)
	/dozzle/internal/docker/log_reader.go:62 +0xfd
github.com/amir20/dozzle/internal/container.(*EventGenerator).consumeReader(0xc000486720)
	/dozzle/internal/container/event_generator.go:85 +0x32
created by github.com/amir20/dozzle/internal/container.NewEventGenerator in goroutine 231
	/dozzle/internal/container/event_generator.go:46 +0x185
{"level":"info","version":"v8.14.12","time":"2025-12-27T22:48:15Z","message":"Dozzle version v8.14.12"}
{"level":"info","version":"v8.14.12","clients":1,"time":"2025-12-27T22:48:16Z","message":"Connected to Docker"}
{"level":"info","version":"v8.14.12","time":"2025-12-27T22:48:16Z","message":"Accepting connections on :8080"}
panic: runtime error: slice bounds out of range [32:27]

goroutine 248 [running]:
github.com/amir20/dozzle/internal/docker.(*LogReader).Read(0xc000cba3f0)
	/dozzle/internal/docker/log_reader.go:62 +0xfd
github.com/amir20/dozzle/internal/container.(*EventGenerator).consumeReader(0xc0009590e0)
	/dozzle/internal/container/event_generator.go:85 +0x32
created by github.com/amir20/dozzle/internal/container.NewEventGenerator in goroutine 244
	/dozzle/internal/container/event_generator.go:46 +0x185

[ivoshm]

I have some news on the topic:

• The good news is that I can reliably simulate the error.
• The bad news is that it involves sensitive data, which I am not able to publish.
• The good news is that I managed to get the Dozzle development environment up and running and I can do some live debugging on it.

The main issue is that Podman sometimes sends log entries without a newline at the end, followed by an extra event containing only an additional timestamp and newline character.

I have verified this behaviour even at the network level.

Example:

2026-03-15T11:53:04Z 15-Mar-2026 11:53:04.957 client @0x7fa652d03c00 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
2026-03-15T11:53:04Z
2026-03-15T11:53:05Z 15-Mar-2026 11:53:05.956 client @0x7fa651e29c00 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
2026-03-15T11:53:05Z

(the first record has a total of 197 characters and last character isn't newline)
(the third record has a total of 207 characters and last character isn't newline)

Unfortunately, these short events only have 22 characters (timestamp, space and newline), and when Dozzle (line 62 of internal/docker/log_reader.go) wants to cut 32 characters from them, it causes panic.

I'm not sure why the continuation of the log message needs at least 32 characters. This is probably a hack for another environment, such as Docker or Kubernetes.

I hope these findings will help resolve this annoying issue for Dozzle users running Podman.
If you need any further information, testing, etc., I'm here to help.

[ivoshm]

It looks like Docker :)

According to https://github.com/moby/moby/blob/10aecb0e652d346130a37e5b4383eca28f594c21/api/server/httputils/write_log_stream.go#L50-L52, it uses the RFC3339NanoFixed format (2026-02-06T04:40:00.679243894Z), which is 30 characters long and includes a space as the 31^(st) character.

EDITED: The Podman team has already "solved" this problem — see containers/podman#27691.
I used version 5.7.0 for testing, but the solution will be available for version 5.8.0.
Unfortunately, however, they use a modified version of RFC3339NanoFixed (https://github.com/containers/podman/blob/8b0f9eb8a4e329826d753bfa6100256305a5a78f/libpod/logs/log.go#L18-L22) that is 35 characters long (timezone included).

Perhaps we could trim continuing log event to the first space character (including it); that would work for both Docker and Podman.

[amir20]

@ivoshm since you have it running locally to debug. Can you try a fix? You can send a PR if it works and I'll review it.

[ivoshm]

Pull request ready ... #4538

[amir20]

Merged. I did some testing. Looks like it hasn't broken anything.