Borg gets SIGKILL'd mid-extraction #267
Labels
No Label
bug
data loss
design finalized
good first issue
new feature area
question / support
security
waiting for response
No Milestone
No Assignees
2 Participants
Notifications
Due Date
No due date set.
Dependencies
No dependencies set.
Reference: borgmatic-collective/borgmatic#267
Loading…
Reference in New Issue
No description provided.
Delete Branch "%!s(<nil>)"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
What I'm trying to do and why
Extract a prior backup from a remote repo using borgmatic.
Steps to reproduce (if a bug)
Not sure if it's a bug.
Actual behavior (if a bug)
Environment
borgmatic version: 1.4.16
borgmatic installation method: pip
Borg version: 1.1.10
Python version: 3.6.8
operating system and version: CentOS 8 / Ubuntu 18.04.3 (yes I tested on 2 different machines in different regions and same)
Yikes! From these logs, I'm not sure why Borg would get SIGKILL'd during
extract
. The first thing that pops to mind though is that perhaps you're running out of memory on the client* duringextract
, and the kernel OOM killer is issuing a SIGKILL to Borg. Can you try looking in your system logs to see if there's any indication of the OOM killer taking Borg out? And how's memory usage look duringextract
?I can't think of anything in borgmatic specifically that would cause that. Do other borgmatic/Borg actions like
create
appear to work okay, and isextract
the only one experiencing a problem?* I'm assuming this is on the client, rather than the server, based on the log output above. But maybe it's worth checking out the memory / OOM situation on the server too just in case I'm wrong about that.
@witten Yes you are right! Just checked the logs and kernel is killing borg. It seems like only
extract
is encountering OOM. I'm running another extract and I will reply to you for the memory usage.What do you think that will be a solution for this issue?
Oh by the way, the server's got 30 GB of RAM.
Is the kernel killing Borg on the client or the server? If on the client, how much free memory is available there at the time
borg extract
runs? If on the server, how much free memory is available there? You can tryfree -h
or similar.One other thing you can do is to try to run the Borg command without borgmatic to see if that exhibits similar behavior. Based on the above, it looks like that'd be:
If this turns out to be a likely Borg-specific issue, then you may have to file a ticket with the Borg project for help. But first I'd recommend seeing how much memory you have available in case you can free some up to make this work.
Thanks for the help!
I've just read a bit of Borg's code here and seems
--progress
is the culprit (the calculation is very slow and takes a lot of memory). Even this is not related to Borgmatic, you offered many help, thank you a lot! Have a nice day :)I'm glad to hear that you've found the likely cause and have a solution for it!