Ad Widget

**MrKen** · 27-03-2009, 11:57

Around 12 months ago I had similar problems running 1.4.2
I have just pulled out my notes from then, which detail what I tried, etc.

So, check these, maybe it may be of assistance to someone. No guarantees though.

1. Increase memory_limit and post_max_size in php.ini to 32M, as described in this thread

We’ll be back soon!

http://www.zabbix.com/forum/showthread.php?t=2876&page=2

but, no change.
2. Delete “#This file is place-holder, bla-bla-bla “ from zabbix.conf.php, as described here

We’ll be back soon!

http://www.zabbix.com/forum/showthread.php?t=4695

my zabbix.conf.php didn’t have those comments, so no change.
3. Change the code in graph.inc.php to pick data from history or trends tables, as detailed here

We’ll be back soon!

http://www.zabbix.com/forum/showthread.php?t=9139

but, no change.
4.
FIXED ??? See http://www.zabbix.com/forum/showthread.php?t=1364

in /var/www/html/zabbix/include/classes/graph.inc.php around line 510 remove max(clock) as clock

$sql="select itemid,round(900*((clock+$z)%($p))/($p),0) as i,count(*) as count,avg(value) as avg,min(value) as min,max(value) as max,max(clock) as clock from history where itemid in ($str) and clock>=".$thi s->from_time." and clock<=".$this->to_time." group by itemid,round(900*((clock+$z)%($p))/($p),0)";

It might help.

MrKen

**alucard95** · 27-03-2009, 16:47

I am running 1.6.2 and the structure of ../include/classes/graph.inc.php changes...

I don't have this line

Code:

$sql="select itemid,round(900*((clock+$z)%($p))/($p),0) as i,count(*) as count,avg(value) as avg,min(value) as min,max(value) as max,max(clock) as clock from history where itemid in ($str) and clock>=".$thi s->from_time." and clock<=".$this->to_time." group by itemid,round(900*((clock+$z)%($p))/($p),0)";

Anybody knows what is the line to change??

**alucard95** · 27-03-2009, 17:09

In fact, I have two graph file:

/var/www/zabbix/include/classes/graph.inc.php

and

/var/www/zabbix/include/graphs.inc.php

In this last file, I found some sql syntax. But I'm not sure about the parts to delete...

Here the file : /var/www/zabbix/include/graphs.inc.php

Attached Files

graphs.inc.php (32.1 KB, 178 views)

**MrKen** · 27-03-2009, 19:32

STOP!

Before you change anything, please keep in mind that what I posted above was what I did to solve this problem in 1.4.2 running CentOS 5. This operation was conducted around 12 months ago!

What I posted was possible suggestions of what has solved this problem for others and what worked for me. When you read the links you will see that while one thing worked for some people, it didn't work for others. There doesn't appear to be a one size fits all.

I believe that the sql syntax has changed since 1.4.2, and therefore looking for a piece of code from 1.4.2 in some 1.6.2 files may well be fruitless.

I should also add that the 1.4.2 Zabbix Server that this affected 12 months ago is still running fine and has had no graphing problems since then.
I also have Zabbix 1.6.2 running on another server, which has never had any gaps in graph problems. Nor have I patched anything related to graphs.

Google is your friend! Try: gaps graphs site:zabbix.com

Cheers

**bbrendon** · 27-03-2009, 20:40

Originally posted by alucard95

Yes, it is exactly the same on all my graphics.

So, with memory_limits=128M in php.ini, I still have the "hole".

So, I give you some precisions about zabbix architecture:

Zabbix application server: Ubuntu server 8.10, IBM xseries 335, 4Go RAM
Zabbix Mysql server: Ubuntu server 8.10, IBM netfinity 6000R, 8Go RAM

Hosts: 172
Items: 2295
triggers: 831
Required server performance, new values per second: 56.376

I check and none of Zabbix server appears overload.

Thanks for your help

Sounds like a mysql problem, though you could try increasing the number of trapper and snmp processes (any maybe others) in zabbix_server.conf as well.

**tchjts1** · 27-03-2009, 23:46

Go to "Administration" --> "Queue" and change the dropdown box to details. You might see some interesting info there.

I have two items that are to be checked on 12/31/1969 - CPU Wait time (avg1) on 2 of my hosts.

The remainder of the items are to be checked more than 3 hours prior to what the current time is. My server date and times are correct, including the hwclock. I also have the correct date.timezone= setting in my php.ini file on the Zabbix app server.

I didn't have that set on the DB server, but that shouldn't matter as Apache runs on the App server, right? Well, I changed it anyway and will see if that resolves it.

Other than that, I have no clue where Zabbix GUI is picking up the incorrect time.

**tchjts1** · 28-03-2009, 02:17

I do believe I found the issue.

There was a name hostname mis-match in one of my zabbix_agentd.conf files.
Why it made EVERY metric for EVERY host fall behind in time is beyond me.

alucard95 -

Check your zabbix_agentd.conf files and make sure the HostName= field matches the name you have for the server in the Zabbix GUI exactly.

After I fixed the mis-match, I stopped and restarted the Zabbix server, all my zabbix agents and MySql.

It appears to be working much better now.

3 days of troubleshooting because of a hostname mis-match. It would be nice for some type of GUI alert for this type of error.

**MrKen** · 28-03-2009, 11:31

Usually when there is a host name mismatch like this, there should be an error message in zabbix_agentd.log like this:

14617:20090328:114530 Can't open jason object

It means that the Hostname in zabbix_agentd.conf does not match a host name in the zabbix database. The name in zabbix_agentd.conf and the hostname in the zabbix frontend must match exactly - case sensitive.

MrKen

**alucard95** · 30-03-2009, 11:18

Originally posted by tchjts1

I do believe I found the issue.

There was a name hostname mis-match in one of my zabbix_agentd.conf files.
Why it made EVERY metric for EVERY host fall behind in time is beyond me.

alucard95 -

Check your zabbix_agentd.conf files and make sure the HostName= field matches the name you have for the server in the Zabbix GUI exactly.

After I fixed the mis-match, I stopped and restarted the Zabbix server, all my zabbix agents and MySql.

It appears to be working much better now.

3 days of troubleshooting because of a hostname mis-match. It would be nice for some type of GUI alert for this type of error.

Hello,

My log file is clean

, on the server and on the agentd.

However, when I check my items ("Administration" --> "Queue"), I have one on 01/01/1970, I can't explain why so, I have disabled this one.

**alucard95** · 08-04-2009, 15:46

always the same problem

Hi all,

I come back because I still have the same problem even worst...

But now, the hole is not the same on all the graph.

The only change since the last time is an upgrade to 1.6.4.

is there an other way to use the graphs or an specific configuration.

Thanks for your help.

**alucard95** · 08-04-2009, 15:53

new information

If I change the time history, for example 6 hours I obtain this:

maybe the problem is with the refresh...

**tchjts1** · 08-04-2009, 17:40

I have the exact same problem as you and have been trying to solve it for going on 2 weeks now, with no luck. (Ubuntu 8.0.4)

And this being a fresh (stock) install of 1.6.4 with no imports and only 2 hosts being monitored - the Zabbix App server and the Zabbix DB server

Administration --> Queue shows there are 80+ items in the "More than 10 minutes" queue.

The problem appears to be that the "next check" (If you click on 'details' for the queue) is more than 2 hours behind the current time.

The date/time on my Zabbix servers is correct, including the hardware clock. My date.timezone in php.ini is correct.

The graphs are just baffling in their rendering behavior. The below screenshots are of the very same graph.

-The first one is a 1 hour view
-The second one to the right is simply a change to one day view
-The thrid one is also the one day view, but clicked on to get full size graph.

Attached Files

**alucard95** · 09-04-2009, 08:16

Hi tchjts1,

I have exactly the same paramter,

a lot of items ( approximatively 600) in more than 10 minutes.

administration -> queue -> detail -> all my items next check is before the current time (between 1 days and a few minutes) as if "next check" was "last check" instead...

**tchjts1** · 09-04-2009, 17:18

Well, since this is a basic install for me, with only the Zabbix servers as hosts, I am going to blow everything away and do another fresh install.

Other people don't seem to have this issue, so must be something we did.

I'll let you know how it turns out.

**tchjts1** · 10-04-2009, 17:33

It appears to be on the MySql side of the setup.
I have separate Zabbix App and DB servers.

To test this, I left the Zabbix server as is and installed a fresh DB on the App server so they are both on the same machine, and everything is working fine.

My original upgrade was from 1.6.1 --> 1.6.2 --> 1.6.3 --> 1.6.4
As far as I know, there was no DB upgrade scripts that were required, other than to add the 3 additional indexes, which I did.

Leading up to the time of the issues, I really didn't have hosts except for the Zabbix servers, I simply did a drop statement on the zabbix DB and then recreated it. That may have been the issue.

Next week I will rebuild the separate DB machine and move back to using that.

Ad Widget

interruption on zabbix server

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment