@CGB Hi, are you running XO from source with root or another user?
If it is another user, it may be not accessible from it

@FagnerMoraes The delta backups are in vdis folder. You can see it just below cache.json.gz

@FagnerMoraes I just realized you are not listing the same folder that what is backed up in your NAS backup job: on the left, you are listing incremental backups from a given folder, but in the logs I analysed the xva are in another folder

@FagnerMoraes Is your backup encrypted?

Moreover, can you show me one metadata file that is supposed to work please?

@FagnerMoraes Hi,
Can you upload your logs please? It is a bit difficult with only screenshots. You can download them from the very modal you are in this screenshot. You can remove any personal name, I just need the logs to see where this comes from but I suspect you have two broken XVA (indeed, you have more than two json metadata).
At first glance, I would advise you to use a new folder in your NAS to be sure everything goes well

@Andrew ok, thank you. This one indeed is not supposed to be triggered indeed.

@Andrew You mean it fails the whole job for this warning? Or this is just a warning inside a successful job?

@Andrew
Your error happens after a successful merge in s3 which removes old alias not found (expected). However, the error given by s3 is not generic so it was not handled by alias checking. This issue was always there but we decided to show the warnings during backup and fix them with this new refactoring.

Can you try this branch please?
https://github.com/vatesfr/xen-orchestra/tree/pbt_fix_warning_backup

@Andrew I think I found the issue, I am working on it

@MajorP93 We are building an automated QA that we keep updating and running regularly (preferably nightly) to add any event possible. We missed this one because we did not took enough failures into account, so we added a regression test for this too.

@Andrew The warning may still appear indeed but I think it is another issue or maybe a good warning to check your backups. Can you provide the new logs please?

An issue was identified in backup jobs running in incremental with commit 100f113 or release 6.5.0. If one of your incremental backup chains was corrupted, it lead to a cleaning of vdis folder resulting into backup lose.

Please update your source at least to 4552fb9 or your XOA to 6.5.1.

If you encountered issues about a warning with aliases, you are concerned. You need to run your backup jobs again to have your backups back on track.

@pilow @acebmxer This should be fixed in 6.5.1 but you will need to run again your backups.

@FagnerMoraes There is now a fix in latest channel now, but you will need to run backup again.

@Andrew @fagnermoraes Hi,
A fix is ongoing here: https://github.com/vatesfr/xen-orchestra/pull/9925

@fagnermoraes can you provide your logs please?

pierrebrunet289 opened this pull request in vatesfr/xen-orchestra

closed fix: move try/catch inside initializations #9925

Hi,
The latest version of XO-lite has an issue, that is why we are temporarily reverting to version 0.19.0. Sorry for the lack of communication on our part.

@jvanabra Hi,
From the first log file you uploaded, I see there is a timeout issue during the transfer. It is this part which triggers a bug (thank you for highlighting that!) where an alias is missing because the directory itself is not present.
There may be two causes to this: either your SR is under a lot of pressure and makes some timeouts, or there is a firewall or network config making timeouts after 72s

@tjkreidl Hi,
You can see how it renders in branch view, part 1 for example: https://github.com/gounthar/Citrix-Tobias-Kreidl-Collection/blob/add-markdown-versions/tale-of-two-servers-part-1-bios.md

@Pilow Hi,
Thank you for your reporting! It should be fixed in the next version we are thinking releasing this month if no surprise happens. There is a big refactoring on the clean part which handles metadata (and size) updates.