Hi we are having a problem and wanted to see if anybody on the forum had advice for us to see if we were on the right track.
We are having the problem described in the following link https://serverfault.com/questions/90...inconsistently
We have the same problem. We lose data sometimes a day at a time across all server.
As you can see the in the text below we currently have 2 problems pollers are
Ack Actions Tags
11:29:47 Average PROBLEM rh_zabbix Zabbix unreachable poller processes more than 75% busy 8h 57m 8s No
12:31:48 Warning PROBLEM rh_zabbix There are items missing data for more than 60 minutes
We seem to have a manageable amount of new values processed per second about 350 on average.
Our database server is not wanting for CPU and it has 24 GB RAM
The zabbix processor has 16 GB RAM.
To solve the above and according to the link above, we believe we have to adjust our settings.
At this time we do not think its the number of database syncers which is the default 4. There did not seem to have any db_syncer issues in the logs.
We mainly want to target increasing the regular pollers, the unreachable pollers and the amount of history cache because according to our stats suggest that those are the processes that are busy.
We did not find any connection problems in the log.
Below are the current configuration. Everything not listed is set to default
StartPollers=32
StartIPMIPollers=3
StartPreprocessors=12
StartPollersUnreachable=2
StartTrappers=8
StartPingers=6
StartDiscoverers=7
StartHTTPPollers=2
CacheSize=1G
CacheUpdateFrequency=180
# StartDBSyncers=4
HistoryCacheSize=32M
TrendCacheSize=32M
Timeout=15
Below are the proposed configuration
StartPollers=42
StartIPMIPollers=3
StartPreprocessors=12
StartPollersUnreachable=8
StartTrappers=8
StartPingers=6
StartDiscoverers=7
StartHTTPPollers=2
CacheSize=1G
CacheUpdateFrequency=180
# StartDBSyncers=4
HistoryCacheSize=128M
TrendCacheSize=32M
Timeout=15
Comment