Zabbix agent can not allocate memory

[lovesa]

Hello,
After upgrading to zabbix 1.8.10 I`ve got such an error:

 12093:20120313:120423.226 cannot allocate shared memory of size 1401072: [28] No space left on device
 12093:20120313:120423.226 cannot allocate shared memory for collector

I'm running Centos 5.6 on OpenVZ VPS;

Zabbix config is simple, only Server specified;

/proc/sys/kernel/shmmax is 13884901888;

free gives:

free
             total       used       free     shared    buffers     cached
Mem:      18432072    5961356   12470716          0          0          0

After starting zabbix I've checked ipcs:

ipcs

------ Shared Memory Segments --------
key        shmid      owner      perms      bytes      nattch     status      
0x0052e2c1 262145     postgres  600        3601301504 140                     

------ Semaphore Arrays --------
key        semid      owner      perms      nsems     
0x7a1741a1 9699328    zabbix    600        9         
0x0052e2c1 6684673    postgres  600        17        
0x0052e2c2 6717442    postgres  600        17        
0x0052e2c3 6750211    postgres  600        17        
0x0052e2c4 6782980    postgres  600        17        
0x0052e2c5 6815749    postgres  600        17        
0x0052e2c6 6848518    postgres  600        17        
0x0052e2c7 6881287    postgres  600        17        
0x0052e2c8 6914056    postgres  600        17        
0x0052e2c9 6946825    postgres  600        17        
0x0052e2ca 6979594    postgres  600        17        
0x0052e2cb 7012363    postgres  600        17        
0x0052e2cc 7045132    postgres  600        17        
0x0052e2cd 7077901    postgres  600        17        
0x0052e2ce 7110670    postgres  600        17        
0x0052e2cf 7143439    postgres  600        17        
0x0052e2d0 7176208    postgres  600        17        
0x0052e2d1 7208977    postgres  600        17        
0x0052e2d2 7241746    postgres  600        17        
0x0052e2d3 7274515    postgres  600        17        
0x0052e2d4 7307284    postgres  600        17        
0x0052e2d5 7340053    postgres  600        17        
0x0052e2d6 7372822    postgres  600        17        
0x0052e2d7 7405591    postgres  600        17        
0x0052e2d8 7438360    postgres  600        17        
0x0052e2d9 7471129    postgres  600        17        
0x0052e2da 7503898    postgres  600        17        
0x0052e2db 7536667    postgres  600        17        
0x0052e2dc 7569436    postgres  600        17        
0x0052e2dd 7602205    postgres  600        17

Tried to ipcrm -s 9699328 and restarting zabbix, but not helped;

Tried increasing shmmax, not helped;

Wired thing, on the hardware node after zabbix start in user_beancounter (vps) increases shmpages failcounts;

Tried increasing shmpages for the vps, not helped;

I also attached strace zabbix_agentd log;

Also have tried solutions from this forum... no result;

[esawyja]

Same problem

Running SLES with Postgres database, I get this as well
3603:20120807:080356.006 Got signal [signal:15(SIGTERM),sender_pid:3507,sender_uid:1104 ,reason:0]. Exiting ...
3507:20120807:080356.006 Zabbix Agent stopped. Zabbix 1.8.3 (revision 13936).
9370:20120807:082730.545 Zabbix Agent started. Zabbix 1.8.3 (revision 13936).
9370:20120807:082730.553 Can't allocate shared memory for collector. [No space left on device]
20752:20120807:090644.497 Zabbix Agent started. Zabbix 1.8.3 (revision 13936).
20752:20120807:090644.500 Can't allocate shared memory for collector. [No space left on device]
On the other boxes with SLES and Oracle it is ok, any suggestions please?

[btriem]

same problem on centos/rhel 6.x

Anyone come up with a solution for this yet?

Thanks!

Brad

[BDiE8VNy]

Such issues can actually only be related to resource limitation either by the hardware node (physical/tunables/UBC) or by the configuration inside the container (tunables/application).

What have you done/configured exactly?