Lior Kaplan wrote:
> Shlomi Fish wrote:
>> Hi all!
>>
>> I had to reboot eskimo twice during the last days. Today, after the
>> second time I rebooted, I decided to investigate and see what caused
>> the problem. So I found the messages like that in /var/log/messages:
>>
>> <<<<
>> Mar 20 19:30:42 eskimo kernel: oom-killer: gfp_mask=0xd2
>> Mar 20 19:31:04 eskimo kernel: DMA per-cpu:
>> Mar 20 19:31:04 eskimo kernel: cpu 0 hot: low 2, high 6, batch 1
>> Mar 20 19:31:04 eskimo kernel: cpu 0 cold: low 0, high 2, batch 1
>> Mar 20 19:31:04 eskimo kernel: Normal per-cpu:
>> Mar 20 19:31:04 eskimo kernel: cpu 0 hot: low 32, high 96, batch 16
>> Mar 20 19:31:04 eskimo kernel: cpu 0 cold: low 0, high 32, batch 16
>> Mar 20 19:31:04 eskimo kernel: HighMem per-cpu: empty
>> Mar 20 19:31:04 eskimo kernel:
>> Mar 20 19:31:16 eskimo kernel: Free pages:        2168kB (0kB HighMem)
>> Mar 20 19:31:16 eskimo kernel: Active:12237 inactive:56638 dirty:0
>> writeback:0 u
>> nstable:0 free:542 slab:3462 mapped:71410 pagetables:1654
>> Mar 20 19:31:16 eskimo kernel: DMA free:1136kB min:28kB low:56kB
>> high:84kB activ
>> e:3488kB inactive:7220kB present:16384kB
>> Mar 20 19:31:16 eskimo kernel: protections[]: 14 284 284
>> Mar 20 19:31:16 eskimo kernel: Normal free:1032kB min:540kB low:1080kB
>> high:1620
>> kB active:45460kB inactive:219332kB present:311280kB
>> Mar 20 19:31:16 eskimo kernel: protections[]: 0 270 270
>>
>> .
>> .
>> .
>> Mar 20 19:31:20 eskimo kernel: oom-killer: gfp_mask=0x1d2
>> Mar 20 19:31:20 eskimo kernel: DMA per-cpu:
>> Mar 20 19:31:20 eskimo kernel: cpu 0 hot: low 2, high 6, batch 1
>> Mar 20 19:31:20 eskimo kernel: cpu 0 cold: low 0, high 2, batch 1
>> Mar 20 19:31:20 eskimo kernel: Normal per-cpu:
>> Mar 20 19:31:20 eskimo kernel: cpu 0 hot: low 32, high 96, batch 16
>> Mar 20 19:31:20 eskimo kernel: cpu 0 cold: low 0, high 32, batch 16
>> Mar 20 19:31:20 eskimo kernel: HighMem per-cpu: empty
>> Mar 20 19:31:20 eskimo kernel:
>> Mar 20 19:31:20 eskimo kernel: Free pages:        1920kB (0kB HighMem)
>> Mar 20 19:31:20 eskimo kernel: Active:48301 inactive:20731 dirty:0
>> writeback:0 u
>> nstable:0 free:480 slab:3524 mapped:71410 pagetables:1654
>> Mar 20 19:31:20 eskimo kernel: DMA free:888kB min:28kB low:56kB
>> high:84kB active
>> :3928kB inactive:3324kB present:16384kB
>> Mar 20 19:31:20 eskimo kernel: protections[]: 14 284 284
>> Mar 20 19:31:20 eskimo kernel: Normal free:1032kB min:540kB low:1080kB
>> high:1620
>> kB active:189276kB inactive:79600kB present:311280kB
>> Mar 20 19:31:20 eskimo kernel: protections[]: 0 270 270
>> Mar 20 19:31:20 eskimo kernel: HighMem free:0kB min:128kB low:256kB
>> high:384kB a
>> ctive:0kB inactive:0kB present:0kB
>> Mar 20 19:31:20 eskimo kernel: protections[]: 0 0 0
>>
>> Googling "oom-killer":
>>
>> http://www.google.com/search?q=oom-killer&ie=UTF-8&oe=UTF-8
>>
>> I see that it has to do with killing processes that occupy memory once
>> the system runs out of such.
>>
>> top after the reboot showed that the RAM was filled in almost
>> entirely, while the swap was empty. Heavy memory occupiers were the
>> apache processes, sympa and mysqld.
>>
>> What should we do about it? 
> 
> Sympa is the only new addition on the server, right ? Can you move the
> lists to beak ? It takes ~20% of memory.
> 
> The load on the apache might come from the move of the mandriva mirror.
> I also suspect rsync to create peaks in memory usage while syncing.

I gave rsync a test, and it doesn't take more than 5% (has to processes,
each take 2.5%. So it's not the one to blame.

-- 

Lior Kaplan
[EMAIL PROTECTED]
http://www.Guides.co.il

Reply via email to