[systemd-devel] Delete stale *.journal-files from containers

[Mantas Mikulėnas]

On Wed, Oct 28, 2015 at 3:39 PM, Peter Paule <systemd-devel at fedux.org>
wrote:

>
> Hi,
>
> I use `docker` to run containers. Each container uses `systemd` as PID 1.
> I pass `-v /var/log/journal:/var/log/journal` to `docker run` to accumulate
> journals on the docker host. Every time a container is started, a new
> journal file is generated based on the machine-id, leaving quite a few
> 8MiB-`system.journal`-files on the system after the container was "stopped".
>
> Example:
>
> ~~~
> docker run --name centos-1 --rm -ti -v /sys/fs/cgroup:/sys/fs/cgroup -v
> /var/log/journal:/var/log/journal feduxorg/centos
> ~~~
>
> Is there way beside `find /var/log/journal -time +30 -delete` to get rid
> of stale old `journal`.files? I tried `MaxRetentionSec=1day` and
> `MaxTimeSec=1day`, but none of this made `systemd-journald` to delete the
> `system.journal`-files.
>

journald doesn't know all possible ways other machines' journals might
appear here – the directory might belong to a *running* container, it might
be written to over NFS by a thin client (where the client's journald might
have different policies), it might be imported by journal-remote (where the
admin might want to keep it for archival purposes), and so on.

Therefore journald will not delete journals with other machine-ids, since
doing so would possibly apply two conflicting policies to the same logs –
yours, and the container's/client's.

-- 
Mantas Mikulėnas <grawity at gmail.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/systemd-devel/attachments/20151029/faa62c64/attachment.html>