I am still beating my head against this issue. I first, migrated all of the
jails off of the jail server (i5-7500, 32GB RAM, 2GB swap), except the jail
running backuppc. It still ran the system out of memory *and* out of swap,
to the point that I had to physically power off and back on. So this
morning, I migrated the backuppc jail to a machine (my desktop) that is
much beefier than the jail server. It sports an i7 (though CPU is not a
problem), 64GB of RAM, and 8GB of swap. I don't understand whatever it is
(presumably on the backuppc jail) is going straight for swap. The desktop
machine had 93% of swap utilized, and still had 50GB of RAM free. It
actually caused my desktop to become unresponsive.
So I rebooted, and ended up manually starting a backup by hand. After
several minutes, swap had gone to zero, and the system still had about 42GB
of free RAM. I'm not sure where the problem lies, but it appears to be on
the jail side of the equation ratiher than the NAS side. Here is what my
desktop currently looks like from top:
22 processes: 1 running, 21 sleeping
CPU: 0.8% user, 0.0% nice, 0.3% system, 0.3% interrupt, 98.7% idle
Mem: 2032M Active, 4706M Inact, 359M Laundry, 4518M Wired, 121M Buf, 51G
Free
ARC: 2104M Total, 1033M MFU, 968M MRU, 3476K Anon, 6837K Header, 92M Other
1774M Compressed, 2900M Uncompressed, 1.64:1 Ratio
Swap: 8192M Total, 4652M Used, 3539M Free, 56% Inuse, 48K In
And on the NAS, which has 32 GB of RAM and 4GB swap:
51 processes: 1 running, 50 sleeping
CPU: 0.0% user, 0.0% nice, 0.6% system, 0.0% interrupt, 99.4% idle
Mem: 149M Active, 1328M Inact, 28G Wired, 1435M Free
ARC: 24G Total, 15G MFU, 8421M MRU, 256K Anon, 203M Header, 817M Other
22G Compressed, 25G Uncompressed, 1.13:1 Ratio
Swap: 4096M Total, 4096M Free
The other side of the equation is the number of transfer errors that are
occurring on (scheduled) backups. The green ones are backups I ran
manually, to try to get the systems to misbehave. So one one side, I am
seeing transfer errors to the NAS (though a zfs scrub is showing zero
errors, but on the backuppc/jail side, the systems are running to at least
swap, and in many cases, memory starvation.
Does anyone have ideas? Is G.W. right and running backuppc over nfs is a
bad idea? Is there something in backuppc that is preferring swap to RAM?
I've been running backuppc for 20 years, and on FreeBSD for 10, and have
never seen this kind of behavior.
Thanks,
--b
_______________________________________________
BackupPC-users mailing list
BackupPC-users@lists.sourceforge.net
List: https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki: https://github.com/backuppc/backuppc/wiki
Project: https://backuppc.github.io/backuppc/
