Buggy Zabbix 1.4.1 agent on AIX

[dsipe]

I try to migrate from existing 1.1.7 production zabbix environment to Zabbix 1.4.1.

I must admit that it's a very very very difficult task!
First, i have to restart everything from a scratch new Zabbix 1.4.1 server: the mysql migration script is to buggy.

Now, i've just compiled the Agent on one of my AIX server [ ./configure --enable-agent ].
It compiled well and started well to.
BUT impossible to monitor items type 'Zabbix agent (active)' : nothing happen!
So i try to modify all my items to put type to 'Zabbix agent': now it work.

Some of my items are type 'Zabbix trapper'.
When issuing a 'zabbix_sender' command on the AIX, i've got the following message :

root@UXDEVT01 /softs/zabbix/bin >zabbix_sender -z 128.128.12.10 -p 10051 -s uxdevt01 -k sys.cpu.sys -o 5
sent: 0; failed: 1; total: 1

As i said it before, Zabbix is really a nice product, but BUGS MUST STOP.
It's not possible to believe that 1.4.1 is a stable version.

And i don't start yet talking about the bugs i encountered on the Windows server....

[vstorez]

Hello,

Perhaps a possible fix :
In your command, the argument of the -s option must be the Zabbix hostname, and not the UNIX hostname :
root@UXDEVT01 /softs/zabbix/bin >zabbix_sender -z 128.128.12.10 -p 10051 -s uxdevt01 -k sys.cpu.sys -o 5

So, uxdevt01 must be the caracters string with which you create the item in the frontend of zabbix server.
Is that the case ?

[dsipe]

yes, that's the case.

[Alexei]

Cool, one "bug" was stopped...

[yvesd]

I have the same problem.

I use the correct parameters but the command take some time before the answer:
./zabbix_sender -v -z zabbixserver -p 10051 -s power51 -k aix.dummy.test -o 32
zabbix_sender [19652]: Warning: Timeout while executing operation
sent: 0; failed: 1; total: 1

Is it a bug or me?

Thanks in advance

[kempkep]

I have the same problem when using zabbix_sender from Solaris to Linux server. There is no problem Linux to Linux.

The sent data does get received by the server even though the timeout occurs and the status indicates a failure.

I snooped/tcpdumped the network traffic and saw the data being sent and an 'ack' packet recieved. Then the timeout period passes, the connection closes, and then the server sends the OK which is ignored.

[kempkep]

The problem is an endian bug. I have created a patch and submitted it to the bugs forum - title "zbxcomms bug for communication between Solaris and Linux "

It fixed my problem, and probably will sort out the AIX one as well.

I also agree that there are a few bugs in 1.4.1, though the ones I'm hitting are mainly due to using 'bleeding edge' features eg: distributed monitoring.

Still, I'm a keen zabbix advocate and will work through the issues. Roll on 1.4.2 !

[Alexei]

FYI This issue is fixed in 1.4.2.