Zabbix Server dies on any mysql-error?

[juice]

This error seems to occur when I add a new host from template, 3 host where added normally, on 4th it's died. After restarting zabbix-server anything works fine.

012633:20070129:140839 Enabling host [billing]
012633:20070129:140840 Parameter [vfs.dev.write[hda,,avg1]] is not supported by agent on host [billing] Old status [0]
012633:20070129:140842 Parameter [vfs.dev.read[]] is not supported by agent on host [billing] Old status [0]
012633:20070129:140845 Query::select 0,lastvalue from functions where functionid=???:???
012633:20070129:140845 Query failed:You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near '???:???' at line 1 [1064]
012613:20070129:140845 One server process died. Shutting down...
012613:20070129:140845 ZABBIX server is down.

Zabbix 1.1.15 on Gentoo 2.6.17, Mysql 5.0.26

My Zabbix shuts down on any error in the zabbix child.
No message is given, zabbix just dies and the monitoring turnes off.
Is it normal? What am I doing wrong?

[bbrendon]

Thats normal. Zabbix server dies when it has a sql problem. However, I can't say what you're doing that causes the program to fail.

[dantheman]

I think this problem with database errors is being addressed in 1.3.x beta versions of zabbix..

[bbrendon]

There are always bugs being worked out, but 1.1.4 and 1.1.5 are considered production and stable.

[Alexei]

Yes, 1.1.x is stable and I'm not aware of any problems in 1.1.x which may cause this. It is really hard to debug without seeing everything (configuration, how it was compiled, full log files, etc etc).

[brandy]

so, what's going on with this bug? I think the zabbix server should reconnect in case of mysql connect failures. i am using 1.1.5 and this is still an issue!

[DiedX]

so, what's going on with this bug? I think the zabbix server should reconnect in case of mysql connect failures. i am using 1.1.5 and this is still an issue!

Rephrase: YOU think that it is an bug/issue. I had this arguement many times (I think that reconnecting every 60 seconds is a nice fix). On the other hand: it is MySQL that fails: should Alexei reprogram zabbix_server because of this?

If you need it that urgently, check for existence of Zabbix, and after that: check MySQL. If both fail: kick It isn't that hard

[James Wells]

Greetings,

On the other hand: it is MySQL that fails: should Alexei reprogram zabbix_server because of this?

Actually, Zabbix shouldn't fail immediately on a MySQL disconnect. There are many causes of a disconnect, and most of them are not related to MySQL at all. In fact there was a patch that was submitted over 12 months ago that resolved this issue, but was rejected by Alexei, without comment. The patch would simply attempt a re-connect any time an update or query failed against the DB. If it failed to reconnect for more than 60 seconds then Zabbix would fie, but it would post to the log file that it was having issues before hand.

[DiedX]

(I think that reconnecting every 60 seconds is a nice fix)

I think that we fully agree

[James Wells]

I think that we fully agree

Not quite. Or perhaps I am mis-reading it, but you are advocating re-connecting every 60 seconds, even if the connection is still good. Insted the patch that was posted and my own personal opinion is that it should keep the existing connection as long as possible, only attempting to re-connect if it should lose connectivity first.