Flushing podman logs

[f1yn]

I have several dozen pods running at any given time. As my usage has grown, I've noticed that the usability of podman logs become unbearable as logs are usually slow to rotate, even when containers.conf has a smaller log size. This means my terminal gets slammed by days (or even weeks) worth of logs.

There's several solution to this for docker, but they don't translate well to podman. How can I manually flush logs in podman, either totally or on a per container basis? Has anyone figured this out?

[rhatdan]

What log driver are you using? I would figure journald would not have this issue. You can also turn off logging, for containers/pods that you don't want the logs for.

[f1yn]

Hey @rhatdan,

I haven't set any other log driver, so I'm assuming it'd just be journald. Unfortunately for auditing purposes (since some of these containers are connected to a VPC), I don't feel comfortable selectively disabling logs on a per-container basis.

It would appear it's journald (like I suspected):

               "LogConfig": {
                    "Type": "journald",
                    "Config": null,
                    "Path": "",
                    "Tag": "",
                    "Size": "0B"
               },

[f1yn]

Is there an explicit difference between configuring the log size via containers.conf vs --log-opt max-size?

[rhatdan]

No idea.
@nalind @Luap99 Any idea?

[f1yn]

Actually, turns out I made a mistake somewhere. I recently did a Fedora Server upgrade, which is around where the problem started.

During the upgrade, it would seem like /usr/share/containers/containers.conf was reset to it's default value. Not ideal, but that would explain why this suddenly started happening.

[f1yn]

Since I run everything rootless, I'll just opt to set the local user containers.conf instead 😂

[f1yn]

Thanks for your time @rhatdan

[rhatdan]

If you want to customize containers.conf system wide, you should do it in /etc/containers/containers.conf