Ad Widget

**tekknokrat** · 02-12-2008, 14:02

I am using zabbix in a productional and lab-environment.
The issue with eating mem. is only happening in the production environment. The lab-environment is virtual. In the production environment the only difference is that I have also added UserParameter checks (beneath others) for reading hd/cpu temperature:

UserParameter=up.system.cpu.comm[*],ps S -C $1 -o pcpu= --sort=-pcpu | head -1
UserParameter=up.sensor[*],sensors|sed -n '/$1/p'|sed 's/$1: *+$[0-9]\{1,3\}\.[0-9]\{1,3\}$.*/\1/'

**bbrendon** · 03-12-2008, 20:49

Originally posted by Alexei

Well, the memory leak is not identified yet. What item type do you use on the proxies? Passive, active, SNMP, anything else?

For the server, I don't use SNMP. Most of the agents are active and most are v1.4.6. I also use web monitoring.

Libs I use: libcurl-gnutls.so.3, libmysqlclient.so.15.

Zabbix 1.6.1 is 100% reliable for me. No memory leaks that I have seen.

**elvar** · 03-12-2008, 22:32

Originally posted by bbrendon

For the server, I don't use SNMP. Most of the agents are active and most are v1.4.6. I also use web monitoring.

Libs I use: libcurl-gnutls.so.3, libmysqlclient.so.15.

Zabbix 1.6.1 is 100% reliable for me. No memory leaks that I have seen.

I'm betting it's SNMP even more than that is the leak issue.

**Alexei** · 03-12-2008, 22:38

We did more testing and one round of partial code review. We found one memory leak related to handling of external checks only. I do not think external checks are used frequently.

It does look like a memory leak in net-snmp libs to me so far.

**teferi** · 03-12-2008, 22:46

Originally posted by elvar

I'm not sure if I am having this or not, but my Zabbix server at a colo is currently unreachable and I just recently upgraded to 1.6.1. Just before becoming unreachable I received Zabbix alerts saying that the swap file was used up.

This is not server's issue. We're discussing zabbix_proxy here

**cioris** · 05-12-2008, 04:51

still having problems...

Hi all,

Sorry for no posting in the last period, but I was in a short vacation...uuhhhh.. it's good to rest from time to time..

.

I was very surprized to see how many other postiongs appeared meanwhile. Initialy I was alone, so I was thinking I was doing something wrong. Good to see there are some others in my situation... at least I know I'm not stupid...

.

My proxy is used just for active checks, and all clients connected to proxy are windows machines. Here is the list of monitored parameters (I hope it helps):

- Automatic Update Service - service_state[wuauserv]
- Free disk space on c: - vfs.fs.size[c:,free]
- host Information - system.uname
- Host uptime - system.uptime
- Physical free memory - vm.memory.size[free]
- Free physical memory % - vm.memory.size[pfree]
- Total physical memory - vm.memory.size[total]
- Ping to Server - agent.ping
- Processor load - system.cpu.util[]
- DHCP client - service_state[Dhcp]
- Terminal Service - service_state[TermService]
- Total disk space on c: - vfs.fs.size[c:,total]
- Processor load 2 - perf_counter[\Processor(_Total)\% Processor Time]
- used space on c: - vfs.fs.size[c:,used]
- agent version - agent.version
- free swap - system.swap.size[,free]
- free swap % - system.swap.size[,pfree]
- Total swap - system.swap.size[,total]
- Vnc Service - service_state[winvnc]

Again, all are active checks. Please let me know if you see anything wrong w/ one or more parameter causing the proxy to fail.

As a temporary solution I used crond to restart the proxy every 24h. Otherwise it crashes after 4-5 days (after it eats all memory and swap).

BR

**k-ren** · 10-12-2008, 14:59

I met the same problem on proxy

**cioris** · 10-12-2008, 23:21

any news? somebody working on this problem?

**k-ren** · 11-12-2008, 07:58

I am using zabbix1.6.1 on Redhat Enterprise5.
The memory leak will happen just like this.

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
6383 zabbix 20 5 11284 2760 1728 S 2 0.5 0:37.52 zabbix_proxy
6384 zabbix 20 5 11284 2760 1728 S 2 0.5 0:37.44 zabbix_proxy
6385 zabbix 20 5 11284 2760 1728 S 2 0.5 0:36.84 zabbix_proxy
6386 zabbix 20 5 11284 2764 1732 S 2 0.5 0:37.60 zabbix_proxy
6378 zabbix 20 5 7988 1644 984 S 0 0.3 0:01.71 zabbix_proxy
6376 zabbix 23 5 7748 824 524 S 0 0.2 0:00.01 zabbix_proxy
6377 zabbix 20 5 17968 11m 976 S 0 2.2 0:02.61 zabbix_proxy
6379 zabbix 20 5 11284 2764 1732 S 0 0.5 0:37.64 zabbix_proxy
6380 zabbix 20 5 11284 2760 1728 S 0 0.5 0:37.22 zabbix_proxy
6381 zabbix 20 5 11284 2760 1728 S 0 0.5 0:37.98 zabbix_proxy
6382 zabbix 20 5 11284 2760 1728 S 0 0.5 0:37.96 zabbix_proxy
6387 zabbix 20 5 11284 2760 1728 S 0 0.5 0:37.90 zabbix_proxy
6388 zabbix 20 5 11284 2764 1732 S 0 0.5 0:37.31 zabbix_proxy
6389 zabbix 30 5 7748 1228 844 S 0 0.2 0:00.00 zabbix_proxy
6390 zabbix 30 5 7748 1228 844 S 0 0.2 0:00.00 zabbix_proxy
6391 zabbix 30 5 7748 1228 844 S 0 0.2 0:00.00 zabbix_proxy
6392 zabbix 30 5 7748 1228 844 S 0 0.2 0:00.00 zabbix_proxy
6393 zabbix 30 5 7748 1228 844 S 0 0.2 0:00.00 zabbix_proxy
6394 zabbix 21 5 7748 1272 884 S 0 0.2 0:00.05 zabbix_proxy
6395 zabbix 20 5 7748 1268 880 S 0 0.2 0:00.00 zabbix_proxy
6396 zabbix 20 5 11124 2388 1496 S 0 0.5 0:00.18 zabbix_proxy
6397 zabbix 20 5 7748 1240 844 S 0 0.2 0:00.93 zabbix_proxy
6398 zabbix 20 5 7748 1240 844 S 0 0.2 0:00.93 zabbix_proxy
6399 zabbix 21 5 7748 1240 844 S 0 0.2 0:00.96 zabbix_proxy
6400 zabbix 20 5 7748 1240 844 S 0 0.2 0:00.96 zabbix_proxy
6403 zabbix 20 5 7748 1240 844 S 0 0.2 0:00.80 zabbix_proxy
6405 zabbix 20 5 11124 2384 1488 S 0 0.5 0:00.10 zabbix_proxy

The memory usage of pid6377 will grow biger and biger.
And in the proxy debug log of pid6377
##################
6377:20081211:141650 Query [SET CHARACTER SET utf8]
6377:20081211:141650 In process_configuration_sync()
6377:20081211:141650 In connect_to_server() [192.168.1.127]:10051 [timeout:600]
6377:20081211:141650 In get_data_from_server() [request

roxy config]
6377:20081211:141650 In send_data_to_server() [{
6377:20081211:141650 In recv_data_from_server()
6377:20081211:141650 Received [{
6377:20081211:141650 In process_proxyconfig()
6377:20081211:141650 Query [begin;]
6377:20081211:141650 In process_proxyconfig_table() [tablename:drules]
6377:20081211:141650 Query [select druleid from drules]
6377:20081211:141650 End process_proxyconfig_table()
6377:20081211:141650 In process_proxyconfig_table() [tablename:dchecks]
6377:20081211:141650 Query [select dcheckid from dchecks]
6377:20081211:141650 End process_proxyconfig_table()
6377:20081211:141650 In process_proxyconfig_table() [tablename:hosts]
6377:20081211:141650 Query [select hostid from hosts]
6377:20081211:141650 Query [update hosts set host='A-testhost01_group1',dns='A-testhost01_group1',useip=0,ip='192.168.1.124',port =10050,status=0,useipmi=0,ipmi_port=623,ipmi_autht ype=-1,ipmi_privilege=2,ipmi_username='',ipmi_password= '' where hostid=10269;
6377:20081211:141650 End process_proxyconfig_table()
6377:20081211:141650 In process_proxyconfig_table() [tablename:items]
6377:20081211:141650 Query [select itemid from items]
6377:20081211:141650 Query [update items set type=1,snmp_community='public',snmp_oid='1.3.6.1.2 .1.2.2.1.10.2',snmp_port=161,hostid=10269,key_='ke yOutCompletekeyNumber_ab_601',delay=60,status=0,va lue_type=3,trapper_hosts='',units='',multiplier=0, delta=2,snmpv3_securityname='',snmpv3_securityleve l=0,snmpv3_authpassphrase='',snmpv3_privpassphrase ='',formula='1',logtimefmt='',templateid=0,valuema pid=0,delay_flex='',params='',ipmi_sensor='' where itemid=28114;
6377:20081211:141651 End process_proxyconfig_table()
6377:20081211:141651 Query [commit;]
6377:20081211:141651 Sleeping for 299 seconds

##################
It means when zabbix proxy get config date from zabbix server and update it's database,the process will eat memory but not free it.
Who can tell me the reason.

**k-ren** · 12-12-2008, 14:36

File:
src/zabbix_proxy/proxyconfig/proxyconfig.c
Function:
process_proxyconfig_table
Code:
new = zbx_malloc(new, new_alloc * sizeof(zbx_uint64_t));
...
old = zbx_malloc(old, old_alloc * sizeof(zbx_uint64_t));
...
sql = zbx_malloc(sql, sql_alloc * sizeof(char));
...

I found that zbx_free is not be used to release the memory.
Is zbx_free neet to be use?
I added zbx_free,but did not resolve the problem.

**Alexei** · 12-12-2008, 15:35

Originally posted by k-ren

File:
src/zabbix_proxy/proxyconfig/proxyconfig.c
Function:
process_proxyconfig_table
Code:
new = zbx_malloc(new, new_alloc * sizeof(zbx_uint64_t));
...
old = zbx_malloc(old, old_alloc * sizeof(zbx_uint64_t));
...
sql = zbx_malloc(sql, sql_alloc * sizeof(char));
...

I found that zbx_free is not be used to release the memory.
Is zbx_free neet to be use?
I added zbx_free,but did not resolve the problem.

The code is absolutely fine. Thanks you for the hint in your previous email, it does look like we have a problem in a process, which takes care of configuration updates.

**k-ren** · 12-12-2008, 16:06

Alexei
I am collecting 2000 snmp items with one proxy.
And the ConfigFrequency is 300 seconds.
The memory usage of ConfigUpdate process grows bigger very quickly.
Could you give me a advice to resove it.
Thank you.

**Alexei** · 12-12-2008, 23:54

We are performing more testing yearly next week. Please expect more news soon.

**Sasha** · 15-12-2008, 13:43

Problem fixed in version pre 1.6.2 rev. 6454. Problem registered in ZABBIX support system as ZBX-628.

**k-ren** · 15-12-2008, 13:56

Originally posted by Sasha

Problem fixed in version pre 1.6.2 rev. 6454. Problem registered in ZABBIX support system as ZBX-628.

Thank you,and where to the zabbix 1.6.2.
I am using 1.6.1 in service, Could you tell me how to fixed this bug on zabbix 1.6.1

Ad Widget

memory leak in zabbix proxy 1.6.1

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment