Re: kswapd causing giant load

From: Gaspar Bakos
Date: Sat Jun 11 2005 - 18:02:40 EST


Hi, Rik,

RE:
> Looks like your system is NUMA, meaning that the system
> memory is divided in half, and the threshold at which a
> memory dirtying process is throttled might be higher than
> the amount of memory available in the zone.
>
> Can you reproduce this problem if you boot with "numa=off"
> or if you "echo 20 > /proc/sys/vm/dirty_ratio" ?

After the "echo 20 > /proc/sys/vm/dirty_ratio", the system load still
goes up to aboe 3.0, with kswapd0 and kswapd1 (plus cp on top of the
list). By the way, i am using cp between to SATA disks, this time,
with the 3ware arrays not involved in the process at all.

---------------
top - 18:50:20 up 7 days, 6:15, 1 user, load average: 3.15, 2.15, 1.14
Tasks: 91 total, 1 running, 89 sleeping, 1 stopped, 0 zombie
Cpu(s): 0.2% us, 11.7% sy, 0.0% ni, 1.5% id, 84.0% wa, 0.8% hi, 1.8% si
Mem: 4011484k total, 3987444k used, 24040k free, 0k buffers
Swap: 7823576k total, 22176k used, 7801400k free, 3823104k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7521 root 18 0 44420 636 548 D 17.3 0.0 0:34.02 cp
174 root 15 0 0 0 0 S 3.0 0.0 20:18.99 kswapd1
175 root 15 0 0 0 0 S 2.3 0.0 33:05.12 kswapd0
7526 root 15 0 0 0 0 D 1.0 0.0 0:00.71 pdflush
----------------

2.6.11-1.14_FC3smp #1 SMP Thu Apr 7 19:36:23 EDT 2005 x86_64 x86_64 x86_64 GNU/Linux

Cheers
Gaspar
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/