A problem after backup (after migration to 4.x)

My YunoHost server

Hardware: computer
YunoHost version: 4.0.8
I have access to my server : Through SSH | through the webadmin
Are you in a special context or did you perform some particular tweaking on your YunoHost instance ? : yes
If yes, please explain: having a reverse proxy in front

Description of my issue

I have successfully migrated from 3 to 4 and my apps seems to be working fine.
I’m running Synapse, Seafile and Baikal

The problem now is that after I have started a back-up from web GUI the yunohost-api goes crazy and loads the CPU to near 100%.
If I restart the yunohost-api I can use the web GUI except for the pack-up part, if I press “Local archives” the problem is back.

The backup seems to be Ok;
$ sudo yunohost backup list
archives:
- 20201114-154909

Any help on how to fault find would be much appreciated!


So the problem is that creating a backup eats all the CPU ? … Imho it’s not entirely unexpected … except maybe if it does eats all the CPU for a loooong time …

Or is it that listing the existing archives eats the CPU ?

Hi Aleks,

I dont think it during the backup, I think it is still eating CPU after the backup is done(?) So it looks like I cant check my local backups via web GUI after the backup is ready. “yunohost” service was not loading the CPU only “yunohost-api”. Normally you get a timeout on web GUI if something is going on (I did not get this) like - “There is already a YunoHost operation running. Please wait for it to finish before running another one.”, right?

I will do a new backup via SSL and check again to be sure.

Any other ideas?

Uh wokay, do you have anything that would confirm that hypothesis ? Like are you able to run the command 'top' to check what’s eating the CPU ?

yeah, I did take two screenshots :slight_smile:

Hmokay

  • can you tell approximately how many backups archive you have (1, 10, 100…)
  • are these some huge backups (>100M or 1G)

There is about 10 archives and 3 of them are big (Seafile), like 300GB.

Im doing a full backup now via SSL and it looks Ok so far, GUI is giving me the message that its occupied…
I will let it finish and then check again (it will take hours).

Ugh yeah alright … I’m not that much surprised that backup operations (even listing the archive) may take a while then … Though it’s still a “bug”, which imho is related to the fact that so far we’re Yunohost created compressed archives (.gz) which therefore require a shitload of CPU for large archives I think - even though that’s probably relevant (because large archives are usually large because of media file which are … already compressed…)

With Yunohost 4.1, archives will be uncompressed by default (regular .tar instead of .tar.gz) which should mitigate that kind of issues

Ok, thanks for the info.

FYI - forgot to tell, in 3.8 this was never causing this issue as far as I know.

lets wait a few hours and see now… :slight_smile:

ups :grimacing:

Info: [########++…] > Backing up fail2ban configuration
Info: [##########++…] > Backing up synapse configuration…
Info: [############++…] > Backing up systemd configuration…
Info: [##############+…] > Backing up synapse data…
Info: [###############++…] > Backing up synapse log…
Info: [#################+…] > Backing up synapse hook…
Info: Collecting files to be backed up for seafile…
Info: [+…] > Loading installation settings…
Info: [#+++…] > Backing up code…
Info: [####++++++++++++…] > Backing up user data…
Info: [################+…] > Backing up configuration…
Info: [#################+…] > Backing up database
Info: Creating a backup archive from the collected files…
Killed

After moving all big Seafile backup files everything is working as expected.
I guess I have to do it manually for Seafile then - https://seafile.gitbook.io/seafile-server-manual/administration/backup-and-recovery.

edit: …and wait for 4.1 :wink:

This topic was automatically closed 15 days after the last reply. New replies are no longer allowed.