Service cyrus-imapd does not start after restore

[Dashamir Hoxha]

On Wed, Jul 3, 2019 at 12:31 PM Jan Kowalsky <jankow at datenkollektiv.net>
wrote:

> Hi Dashamir,
>
> sorry, it took some time for answering. Did you got any success meanwhile?
>
> Am 27.06.19 um 18:40 schrieb Dashamir Hoxha:
> > On Thu, Jun 27, 2019 at 3:11 PM Dashamir Hoxha <dashohoxha at gmail.com>
> wrote:
> >
>
> >>
> >> The only messages that I can find on `/var/log/mail.log` are these
> >> (immediately after I try `systemctl start cyrus-imapd`):
> >> ```
> >> Jun 27 09:58:12 kolab ctl_cyrusdb[767]: skiplist: clean shutdown file
> >> missing, updating recovery stamp
> >> Jun 27 09:58:12 kolab ctl_cyrusdb[767]: recovering cyrus databases
> >> Jun 27 09:58:12 kolab ctl_cyrusdb[767]: done recovering cyrus databases
> >> Jun 27 09:58:12 kolab master[766]: process type:START name:idled
> >> path:/usr/lib/cyrus-imapd/idled age:0.000s pid:768 exited, status 1
> >> Jun 27 09:58:12 kolab master[766]: can't run startup
> >> Jun 27 09:58:12 kolab master[766]: exiting
> >> ```
>
> it looks like a problem with idled process.
>
> could you run cyrus from command line in debugging mode?
>
> cyrus-master -D
>
> (https://linux.die.net/man/8/cyrus-master)
>
> >>
> >>> As long we haven't more information it's hard to guess. One point which
> >>> comes in my mind are file access rights for cyrus. Since cyrus is
> >>> running as user cyrus the databases and the mailspool needs the rights
> >>> set accordingly.
> >>>
> >>
> >> The ownership seems to be Ok (cyrus:mail) on `/var/spool/imap` and
> >> `/var/lib/imap`.
> >> The restore script also takes care of them:
> >>
> https://gitlab.com/docker-scripts/kolab/blob/master/scripts/restore.sh#L47
> >>
> >> If you had access to the server, would you be willing to have a look an
> >> try to find where is the problem?
> >> I can install sshd on it and give you access.
>
> If the problem still exists: Please try to send some debugging
> information. But, if nothing helps, I could. But I have very little
> experience with docker and do not feel very comfortable without direct
> access to the cyrus instance ;-). Or is it a single container with full
> stack and ssh?
>
> good luck and kind regard.
>

I did not have any luck yet.

The container is a full stack system, with systemd and everything else
(including ssh), so you don't need any docker experience and you will have
direct access to everything.

The problem is that I am currently trying another solution (mailcow). As
soon as I am done with this I may have another look at kolab (it is not
possible to run to mail servers in parallel since they need to use the same
ports). If I can't fix the problem again, I may ask again for help.

However, if someone has time to have a look at it, the scripts for building
the container are here: https://gitlab.com/docker-scripts/kolab
You can reproduce the problem by making first a backup (with `ds backup`),
then rebuild the container (with `ds make`), then install kolab again
(`kolab-setup`), then restore the backup (with `ds restore
backup-file.tgz`).

Thank you for trying to help anyway.
Best regards,
Dashamir



>
> Jan
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kolab.org/pipermail/users/attachments/20190703/8374ea19/attachment.html>