podman-remote with journald+file mishmash: crashes client

[edsantiago]

With this containers.conf:

[containers]
log_driver = "journald"

[engine]
events_logger = "file"

...and this server command:

$ bin/podman --log-level=info system service --timeout=0

...the following sequence yields an unexpected client failure:

$ `bin/podman-remote run -d --name foo quay.io/libpod/testimage:20210610 sleep 30
39e6c6ac379b066642304f6b8d1b7ca12e36fa4023e7bf2061f8e3758ed2fc96
$ bin/podman-remote logs -f foo
Error: channel "123" found, 0-3 supported: lost synchronization with multiplexed stream

Expected: the new "file + journald not supported" error. (Which is shown on the server, but only as info level):

INFO[0016] Request Failed(Internal Server Error): failed to obtain logs for Container 'foo': using --follow with the journald --log-driver but without the journald --events-backend (file) is not supported

[edsantiago]

Oops, forgot to reference #11795

[rhatdan]

#12130 was an attempt to fix this, but I think we should just fail outright if you setup this environment. IE If you set journald for logging, we should require event-logger to use it and vice versa.

[cevich]

Any chance this is related to issue 12029 as well? (I'm grasping at straws here)

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[edsantiago]

Problem still present in main @ 18e3193

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[rhatdan]

@edsantiago we still got this one?

[edsantiago]

Still present in main @ 46d9a25

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[robbiemu]

@cevich it looks possible to me as well. "--events-backend=file" would have specified where to write the logs by default, fi the logs command did not get updated as well, this could be the issue

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[edsantiago]

Yikes - seen just now in a random PR in the wild, possibly a flake

[+0120s] not ok 74 podman logs - basic test
         # # podman-remote [...] create quay.io/libpod/testimage:20220615 echo GSwcgzUZbhgjKIOCdvNIXEVvc7NqybRHOBGb9D2L
         # 0a167f24f01f348ab9f68e24f9e7f714dd90a70dc3155be488e9c929ce4911f0
         # # podman-remote [....] logs 0a167f24f01f348ab9f68e24f9e7f714dd90a70dc3155be488e9c929ce4911f0
         # Error: channel "123" found, 0-3 supported: lost synchronization with multiplexed stream
         # [ rc=125 (** EXPECTED 0 **) ]

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[edsantiago]

Still present in main @ 07c0dd1 (which includes #15717, the multiple-events-problems PR). We're approaching a year of this, and it seems like a non-issue. The June 27 flake bothers me though.

[vrothberg]

Tested with Podman v4.5 and both, local and remote, work.

[Luap99]

I would say I fixed it in #18282 but this is not in v4.5? Did you tested main instead?

[vrothberg]

I am fairly confident of having tested v4.5. Does it not work on your end with 4.5?

[Luap99]

No it fails with v4.5, did you forget to add -f to logs? This is important to trigger the error condition.
You can also try

$ podman-remote logs notexists
Error: channel "123" found, 0-3 supported: lost synchronization with multiplexed stream

Anyway doesn't matter it is 100% fixed on main :)

[vrothberg]

I did use -f so I must have used main. Thanks for clarifying!