VM shutting down

T

Tipika

New Member
Jul 22, 2009
21
0
1
I have a core i7 server with 8gb ram and proxmox 1.5
I have 2 vm, one has 5gb ram, the other one has 1gb ram.
The memory is never full, this is the memory graph over time:
immaginek.gif


Every night i get the 5gb vm turned off.
I found this in the log:
Code: 
Nov 24 09:30:13 server kernel: Call Trace:
Nov 24 09:30:13 server kernel:  [<ffffffff8029edd7>] oom_kill_process+0x87/0x140
Nov 24 09:30:13 server kernel:  [<ffffffff8029f2ff>] out_of_memory+0x14f/0x2c0
Nov 24 09:30:13 server kernel:  [<ffffffff802a359e>] __alloc_pages+0x34e/0x470
Nov 24 09:30:13 server kernel:  [<ffffffff802a56ce>] __do_page_cache_readahead+0x12e/0x290
Nov 24 09:30:13 server kernel:  [<ffffffff802a5c65>] ondemand_readahead+0x1a5/0x230
Nov 24 09:30:13 server kernel:  [<ffffffff8029c7fa>] do_generic_mapping_read+0x14a/0x410
Nov 24 09:30:13 server kernel:  [<ffffffff8029b3e0>] file_read_actor+0x0/0x190
Nov 24 09:30:13 server kernel:  [<ffffffff8029e2b6>] generic_file_aio_read+0x116/0x1d0
Nov 24 09:30:13 server kernel:  [<ffffffff802d1e73>] do_sync_read+0xe3/0x130
Nov 24 09:30:13 server kernel:  [<ffffffff8025c220>] autoremove_wake_function+0x0/0x30
Nov 24 09:30:13 server kernel:  [<ffffffff80251271>] group_send_sig_info+0x91/0x130
Nov 24 09:30:13 server kernel:  [<ffffffff8026165a>] getnstimeofday+0x3a/0xb0
Nov 24 09:30:13 server kernel:  [<ffffffff80260042>] ktime_get_ts+0x22/0x60
Nov 24 09:30:13 server kernel:  [<ffffffff802d2ec8>] vfs_read+0xc8/0x180
Nov 24 09:30:13 server kernel:  [<ffffffff802d312b>] sys_pread64+0x1ab/0x1c0
Nov 24 09:30:13 server kernel:  [<ffffffff8020c69e>] system_call+0x7e/0x83
Nov 24 09:30:13 server kernel: Active:1567052 inactive:400035 dirty:689 writeback:0 unstable:0
Nov 24 09:30:13 server kernel:  free:11722 slab:34626 mapped:2941 pagetables:4830 bounce:0
Nov 24 09:30:13 server kernel: Node 0 DMA free:10440kB min:12kB low:12kB high:16kB active:0kB inactive:0kB present:9936kB pages_scanned:0 all_unreclaimable? yes
Nov 24 09:30:13 server kernel: lowmem_reserve[]: 0 2983 8023 8023
Nov 24 09:30:13 server kernel: Node 0 DMA32 free:24284kB min:4260kB low:5324kB high:6388kB active:2244484kB inactive:711964kB present:3055360kB pages_scanned:874 all_unreclai$
Nov 24 09:30:13 server kernel: lowmem_reserve[]: 0 0 5040 5040
Nov 24 09:30:13 server kernel: Node 0 Normal free:12164kB min:7196kB low:8992kB high:10792kB active:4023724kB inactive:888176kB present:5160960kB pages_scanned:1520 all_unrec$
Nov 24 09:30:13 server kernel: lowmem_reserve[]: 0 0 0 0
Nov 24 09:30:13 server kernel: Node 0 DMA: 4*4kB 7*8kB 6*16kB 3*32kB 3*64kB 6*128kB 2*256kB 1*512kB 0*1024kB 0*2048kB 2*4096kB = 10440kB
Nov 24 09:30:13 server kernel: Node 0 DMA32: 442*4kB 481*8kB 207*16kB 65*32kB 22*64kB 11*128kB 3*256kB 3*512kB 0*1024kB 2*2048kB 1*4096kB = 24320kB
Nov 24 09:30:13 server kernel: Node 0 Normal: 261*4kB 571*8kB 60*16kB 8*32kB 5*64kB 2*128kB 3*256kB 0*512kB 0*1024kB 0*2048kB 1*4096kB = 12268kB
Nov 24 09:30:13 server kernel: Swap cache: add 1396753, delete 1174137, find 12062886/12133475, race 0+80+2
Nov 24 09:30:13 server kernel: Free swap  = 0kB
Nov 24 09:30:13 server kernel: Total swap = 1051056kB
Nov 24 09:30:13 server kernel: Free swap:            0kB
Nov 24 09:30:13 server kernel: 2359296 pages of RAM
Nov 24 09:30:13 server kernel: 319297 reserved pages
Nov 24 09:30:13 server kernel: 447073 pages shared
Nov 24 09:30:13 server kernel: 222616 pages swap cached
Nov 24 09:30:13 server kernel: OOM killed process kvm (pid=11428, ve=0) exited, free=1205790 gen=4.

What's wrong?
 
Do you know if this update is safe for the ovh version with mdadm and softraid 1?
The last time that i tried to update from 1.2 to 1.4 the server was unbootable.
I posted 3 times in the ovh forum, they did not give me a reply...
 
no, don´t know you. and keep in mind, we do never recommend and support softraid (we do no tests with softraid here)
 
I know, but it's not so common to find a dedicated server with hw raid. It would be at least 4 times more expensive.
 
Also OVH offers hardware raid for just around € 20 per month, but its just a basic one. Good controller are more expensive, yes.

but if you think its cheaper to fiddle around with softraid and running an unsupported configuration - just do it but be prepared for some issues like you had already.
But I assume for most businesses this is no option and at the end it is better to invest in good hardware.

the future: in a few years we will have stable mainstream SSD and maybe the overall reliability with these (single) SSD is better than what we have now with raid - just think of the upcoming new Intel SSD with some kind of integrated cache protection in the case of power loss.
 
I updated to 1.6 with kernel 2.6.24 (other kernels are not bootable on my server, says that /lib/modules/2.6.34-xxxx-std-ipv6-64/ was missing and could not continue) but i still get the problem when too much ram is used.

I assigned 7.2 gb out of 8gb to VMs, and then, VM101 was shutted down...
Maybe it's KSM fault? I see with pveversion -v that ksm it's still there

edit: this is the memory graph:
 
Last edited:
we never had a 2.6.34 kernel so where do you get this?
 
i read that message when installing proxmox-ve-2.6.32 via aptitude

edit: more details:
my /etc/apt/sources.list:
deb ftp://mir1.ovh.net/debian lenny main contrib non-free

# PVE packages provided by proxmox.com
deb ftp://download.proxmox.com/debian lenny pve

# security updates
deb http://security.debian.org/ lenny/updates main
Click to expand...
result of pveversion -v
pve-manager: 1.6-5 (pve-manager/1.6/5261)
running kernel: 2.6.24-12-pve
proxmox-ve-2.6.24: 1.6-26
pve-kernel-2.6.24-10-pve: 2.6.24-21
pve-kernel-2.6.24-12-pve: 2.6.24-25
qemu-server: 1.1-22
pve-firmware: 1.0-9
libpve-storage-perl: 1.0-14
vncterm: 0.9-2
vzctl: 3.0.24-1pve4
vzdump: 1.2-8
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1dso1
pve-qemu-kvm: 0.12.5-2
Click to expand...

edit2: i noticed that ksm disappeared - i am 100% sure that yesterday was there after updating but before rebooting
 
Last edited:
You must log in or register to reply here.
Top Bottom