strange graph in master node, anyone can explain?

[webdisk]

I've two zabbix servers deployed in distribute mode, one master, one child. I'm using the master node as monitoring report server...

When I check the graph from master node, I found a server's cpu utilization (monitored by child node) is very strange...

Seem the master node does not calculate the graph correctly.

note the cpu min and max value.

Thanks

[welkin]

i have the same problem, but i haven't found a solution yet.

see: http://www.zabbix.com/forum/showthread.php?t=12642


regards
welkin

[webdisk]

master node donot calculate child's graph over 4 days correctly

Seem this error is related to the graph calculation:

1. for graph <48hrs, zabbix calculate it using history;
2. if >48hrs, zabbix use trend;
3. Page196@ZABBIX Manual v1.6 Release 017 state "Trends are calculated locally based on received historical data."

So the possible cause for the error is due to master node not calculate/store/keep correct trend for graph.

Scaned forum found several thread but seem there is no fix.

Seem version 1.46 worked in the past, I'm using 1.6.4...
refer url:

http://www.zabbix.com/forum/showthread.php?t=10956&highlight=child+node+trend+master

I'm trying to build DM model with one master reporting server facing operator while child node doing actual monitoirng checks (also keep one copy of history/trend/event for redundancy).

So far if child graph function not presented well at master node, then it's not possible to do centralized report model with zabbix.

As this issue impacts overall monitoring deployment, appreciate ZABBIX team can drop advice on this.

Thanks in advance