zabbix_server 1.6 dies periodically

[milprog]

I have zabbix-postgresql 1.6 i386 running on centos 5.2 . Every now and then the server stops, and I don't know why. When I set the errorlevel to debug, I get the messages below, but I have no idea what causes the server to stop. When I restart the server, it runs again for a few minutes to a few hours but then stops again.

What can I do to troubleshoot this? Any help would much be appreciated.

Regards
--Marcel Gsteiger

Here are the last lines of my log:

26273:20081006:092630 Query [insert into history (clock,itemid,value) values (1223277985,21792,0.010000)]
26278:20081006:092630 ItemId:20908 Evaluating last(0)
26278:20081006:092630 In evaluate_function(last)
26278:20081006:092630 End of evaluate_function(result:15361852)
26278:20081006:092630 Do not update functions, same value
26278:20081006:092630 ItemId:20908 Evaluating nodata(900)
26278:20081006:092630 In evaluate_function(nodata)
26278:20081006:092630 In evaluate_NODATA()
26278:20081006:092630 End of evaluate_NODATA()
26278:20081006:092630 End of evaluate_function(result:0)
26278:20081006:092630 Do not update functions, same value
26278:20081006:092630 End update_functions()
26278:20081006:092630 In update_triggers [itemid:20908]
26278:20081006:092630 Query [select distinct t.triggerid,t.expression,t.description,t.url,t.com ments,t.status,t.value,t.priority,t.type from triggers t,functions f,items i where i.s
26273:20081006:092630 In add_trend()
26273:20081006:092630 Query [select num,value_min,value_avg,value_max from trends where itemid=21792 and clock=1223276400]
26272:20081006:092630 End substitute_functions() [1>10]
26272:20081006:092630 In evaluate(1>10)
26272:20081006:092630 In evaluate_simple(1>10)
26272:20081006:092630 In evaluate_simple(1)
26272:20081006:092630 In evaluate_simple(10)
26272:20081006:092630 End evaluate(result:0.000000)
26272:20081006:092630 End evaluate_expression(result:0)
26272:20081006:092630 In update_trigger_value(triggerid:13613,old:0,new:0,1 223277990)
26272:20081006:092630 End update_trigger_value()
26272:20081006:092630 End update_triggers [19736]
26272:20081006:092630 Query [commit;]
26265:20081006:092630 One child process died. Exiting ...
26278:20081006:092630 In evaluate_expression({14965}<600)
26278:20081006:092630 In substitute_simple_macros()
26272:20081006:092630 Got signal. Exiting ...
26278:20081006:092630 In substitute_simple_macros (data:{14965}<600)
26278:20081006:092630 Query [select expression from triggers where triggerid=0]
26273:20081006:092630 Query [update trends set num=78, value_min=0.000000, value_avg=0.023426, value_max=0.050000 where itemid=21792 and clock=1223276400]
26273:20081006:092630 End of add_history
26273:20081006:092630 In update_item()
26273:20081006:092630 In calculate_item_nextcheck (21792,15,,1223277985)
26273:20081006:092630 End calculate_item_nextcheck (result:1223278000)
26273:20081006:092630 Query [update items set nextcheck=1223278000,prevvalue=lastvalue,lastvalue ='0.010000',lastclock=1223277985 where itemid=21792]
26278:20081006:092630 End substitute_simple_macros (result:{14965}<600)
26278:20081006:092630 In substitute_functions({14965}<600)
26278:20081006:092630 Query [select 0,lastvalue from functions where functionid=14965]
26273:20081006:092630 Got signal. Exiting ...
26275:20081006:092630 Got signal. Exiting ...
26278:20081006:092630 Got signal. Exiting ...
26280:20081006:092630 Got signal. Exiting ...
26279:20081006:092630 Got signal. Exiting ...
26281:20081006:092630 Got signal. Exiting ...
26282:20081006:092630 Got signal. Exiting ...
26277:20081006:092630 Got signal. Exiting ...
26283:20081006:092630 Got signal. Exiting ...
26276:20081006:092630 Got signal. Exiting ...
26284:20081006:092630 Got signal. Exiting ...
26286:20081006:092630 Got signal. Exiting ...
26285:20081006:092630 Got signal. Exiting ...
26265:20081006:092632 Query [select oid from pg_type where typname = 'bytea']
26265:20081006:092632 ZABBIX Server stopped. ZABBIX 1.6.

[Alexei]

Please can you try the pre-1.6.1 from www.zabbix.com/developers.php. Let me know if the problem won't disappear.

Thank you.

[milprog]

thank you for your suggestion, I will try that.

Meanwhile, I found out that the problem occurs when storing duplicate values in the history table. Because I have other software that generates reports from the database, I had declared a primary key on this table (the index now being unique). I just thought "how can anyone interpret duplicate values for the same hostid/timestamp pair". Obviously this leads to problems.

I still don't understand how these duplicate values can occur (and why it makes sense to store them in the database), but removing the unique index seemed to solve the problem.

Regards
--Marcel

[Alexei]

I think no one can help because you use a modified database schema. You should have mentioned it beforehand.