Items update interval increased dramatically

[Lufa]

Hi!

When my Zabbix reached ~650 hosts (mostly cable modems with 5-10 items) strange problem appear with items update interval.

Instead of getting item data once a 60s (modems signal level) or once a 10s (e.q. server's CPU load ) it has the data with 20-30 minutes interval.

I thought it might caused by MySQL db (30GB). I have reinstalled Zabbix (1.6.3) completely (all history was deleted) and import all hosts and templates again.

Now, I am getting some data with specified in item properties interval (60s, correctly), but (probalbly) all items still has 20-30 minutes, instead of seconds (mostly: FreeBSD servers with standart FreeBSD teplate - 90% items).

What can be wrong? How can I make a further investigation of this problem?

Thank you!

[Lufa]

Zabbix status (from dashboard) attached.

[Lufa]

epsilon: Processor load (should update every 30s):

2009-04-01 10:45:45 1238571945 0.3896
2009-04-01 10:29:07 1238570947 0.3188
2009-04-01 10:13:16 1238569996 0.5415
2009-04-01 09:54:09 1238568849 1.2842
2009-04-01 09:36:11 1238567771 0.6479
2009-04-01 09:18:41 1238566721 0.7661
2009-04-01 09:08:13 1238566093 0.4204
2009-04-01 08:47:12 1238564832 0.4688
2009-04-01 08:28:08 1238563688 0.3770
2009-04-01 08:08:53 1238562533 0.2144
2009-04-01 07:44:06 1238561046 0.2739
2009-04-01 07:25:45 1238559945 0.1133
2009-04-01 07:04:59 1238558699 0.2695
2009-04-01 06:40:16 1238557216 0.2095

[Lufa]

1.6.4 - the same

I have just update Zabbix from nightly build - still the same problem.

[richardkraal]

I've got the same problem


the queue is also behind schema

ZABBIX server is running Yes -  
Number of hosts (monitored/not monitored/templates) 318 226 / 54 / 38 
Number of items (monitored/disabled/not supported) 13111 8653 / 661 / 3797 
Number of triggers (enabled/disabled)[true/unknown/false] 3577 3232 / 345  [49 / 202 / 2981] 
Number of users (online) 10 2 
Required server performance, new values per second 26.3922 - 

interval is 5min / 300sec


2009.Apr.02 14:38:35 773186.9080
2009.Apr.02 14:26:59 652853.6078
2009.Apr.02 14:13:23 744376.2061
2009.Apr.02 14:02:23 878012.2998
2009.Apr.02 13:53:16 882118.8203
2009.Apr.02 13:43:26 861063.7435
2009.Apr.02 13:32:31 1031574.7059
2009.Apr.02 13:23:27 1595402.0530
2009.Apr.02 13:13:23 1912943.4074
2009.Apr.02 12:59:26 1413415.1282
2009.Apr.02 12:49:02 904657.3827
2009.Apr.02 12:38:14 2385137.0924
2009.Apr.02 12:26:09 973939.3854
2009.Apr.02 12:13:47 701427.9365

[Lufa]

I have decreased number of available Items to ~ 4000 and it just start working well.

When I enabled all 6500 items - updates interval grew to 20-30 min.

P.S.
Most items have 15 min interval in the properties, but the quantity of this items has an influence on items with short interval.

[Alexei]

What can be wrong?

Firewall settings? Other network related issues causing drops of TCP connections? Just a guess...

[Lufa]

Firewall settings? Other network related issues causing drops of TCP connections? Just a guess...

I did not change firewall during last days. But who knows...

80% of items are ICMP requests. But I can not find any network problems, testing the connections intensively.

[Alexei]

80% of items are ICMP requests.

In this case you should probably run more ICMP pingers, parameter StartPingers in zabbix_server.conf.

[Lufa]

In this case you should probably run more ICMP pingers, parameter StartPingers in zabbix_server.conf.

Oh... I did not pay attention to this processes. I thought it is all about simple ping.

I will try!

[Lufa]

In this case you should probably run more ICMP pingers, parameter StartPingers in zabbix_server.conf.

I am very sorry. I had a very long day yesterday.

I mean SNMP items, not ICMP.