onlyjob
I also thought about "non-empty but not valid ServerId" but issue cleared only after manipulation with groups. Before that I did re-save configuration of all cameras, trying to set server...
Thanks for acknowledgement, @chogata. No worries. ;)
Thanks. I think recovery of a missing chunk could be considered as manual action, hence "chunk marked as readable after repair" status should be cleared automatically, upon re-discovery of missing...
Definitely _ext4_. I recommend to avoid _xfs_. Also native _ext4_ encryption is faster than _cryptdisk_, if you want your chunks encrypted.
I specifically tested performance of _ext4_ versus _xfs_ on identical disks on the same chunkserver. _ext4_ is faster on metadata operations and it uses available space better than _xfs_. Anecdotal...
That was not an `rsync` issue; _xfs_ really does consume significantly more space than _ext4_ by default (and that seems to depend on number of files in file system). Maybe...
Situation regarding power-loss reliability may have improved - I do not know. I was saying about my lack of confidence in the file system whose developers cared so little about...
How many files have you created? When only 1% of the file system is used there will be hardly any difference in amount of available space... I've used ~4 TiB...
> what OS version you tested on Debian "Stretch" on amd64 (a.k.a. x86_64). > I would say that this is not XFS specific. Denial detected. :) How could it be...
What bothers you, specifically, in Theodore's interview?