A couple of questions regarding Real-Time Backup Server

I’m running a seafile server and a real-time backup server, using a crontab script to dump the databases every hour and sync them to the backup server.

Everything seems to be working fine except:

The backup server has some kind of issue with a specific file - I’m tailing seafile.log and after any sync I have an error message for a specific file:

[04/11/2019 01:26:49 PM] ../common/fs-mgr.c(1314): [fs mgr] Failed to read file 12r5684dqw34ae48r27c7f841qk6d447609bbt33g.
[04/11/2019 01:26:49 PM] ../common/fs-mgr.c(2370): [fs mgr] Failed to find file 12r5684dqw34ae48r27c7f841qk6d447609bbt33g.

I’ve run seaf-fsck on the main server and it didn’t found any issue; when I manually sync the library using /seaf-backup-cmd.sh it syncs successfully but the error keeps showing up next time there is a change - always the same file id.

  • can I run seaf-fsck on the backup server to check for any errors? I assume I would need to import the current mysql dump before doing it but not sure if it is possible and/or if it will fix anything? how can i solve the issue otherwise?

I have imported one mysql dump into the real-time backup server to test that everything was working using the instructions in the manual found here:

I used the command

mysql -u <user> -p<pass> < dbdump.sql

The import was apparently successful and I just restarted the backup server without any other change (so it continues operating as a real-time backup server).

  • If there is a future failure and I need to import again the latest dump, can I just use the same command and it will overwrite the previous import and update the databases to the latest version? Or do i need to clear the databases from the previous import before attempting to do it again?

Thank in advance for your help.

bumping… anyone?

Hi Fred,

  1. can I run seaf-fsck on backup server?

yes. it is possible to run seaf-fsck on the real-time backup server. Stop all seafile-services and then just run the seaf-fsck.sh.

  1. can I reimport the dump?

yes. There is no problem in reimport the dbdump.sql. If you exported the database like described in the manual then your dump will look like:

CREATE TABLE `Branch` ( ...

Das heißt, die existierenden Tabelle (mit Ausnahme von Repo, RepoHead und Branch) werden gedroppt und dann mit den aktuellen Werten gefüllt.

  1. general

I would try to get to know what kind of file that is. That should help to determine the source of the problem.

Best regards

Hi Cristoph,

Thank you very much for your help.

I’ve looked into one of the sql dumps and it does have the configuration as you posted; I’ve shut down seafile service, imported the latest dump and run a seaf-fsck on the backup server - everything went through smoothly and no errors detected. I’m thrilled to have a proper working seafile set-up with real-time backup server.

Any way to find what file it refers to by its id?

Many thanks once again.