Server and proxy = 7.0.12, but this has been an issue for every version of 7 I've used.
One proxy always has a queue of 60-100K values but I see no reason why.
We have a couple dozen proxies, almost exclusively monitoring network gear via SNMP queries, no traps.
It's also the largest proxy, with 1000 VPS.
But the 2nd place proxy is still 680 VPS, and has a queue of zero. The entire server queue is from this one problem proxy.
But looking at every stat I can think of, I just don't see anything that should limit us.
Using a fraction of CPU/memory/disk IO on the proxy.
Server recently upgraded and has no issues.
I know we're limited to 1000 items per second as the MaxConcurrentChecksPerPoller, but I've forked plenty of pollers, probably too many, but nothing seems to help.
Unless 1000 is a hard limit someplace else on a proxy?
I also know we're using the old version 6 type sync SNMP templates, not the newer v 7 async templates. That's the only problem I know of.
Can I fix my problem by throwing money at it and putting a 2nd proxy at this site and a proxy group?
Proxy settings:
CacheSize=256M
HistoryCacheSize=64M
HistoryIndexCacheSize=32M
ProxyMemoryBufferSize=128M
StartVMwareCollectors=1
# we don't use any vmware at this facility, but I have this as standard on all our proxies
VMwareFrequency=60
VMwarePerfFrequency=60
VMwareCacheSize=16M
ProxyOfflineBuffer=48
StartDiscoverers=3
StartPollers=20
StartSNMPPollers=20
StartPingers=6
StartPreprocessors=6
StartDBSyncers=6
Timeout=3
StartPollersUnreachable=3
MaxConcurrentChecksPerPoller=1000
Graphs below show some atypical activity levels in the last 30 minutes as I've been monkeying with settings (again) to improve performance.
One proxy always has a queue of 60-100K values but I see no reason why.
We have a couple dozen proxies, almost exclusively monitoring network gear via SNMP queries, no traps.
It's also the largest proxy, with 1000 VPS.
But the 2nd place proxy is still 680 VPS, and has a queue of zero. The entire server queue is from this one problem proxy.
But looking at every stat I can think of, I just don't see anything that should limit us.
Using a fraction of CPU/memory/disk IO on the proxy.
Server recently upgraded and has no issues.
I know we're limited to 1000 items per second as the MaxConcurrentChecksPerPoller, but I've forked plenty of pollers, probably too many, but nothing seems to help.
Unless 1000 is a hard limit someplace else on a proxy?
I also know we're using the old version 6 type sync SNMP templates, not the newer v 7 async templates. That's the only problem I know of.
Can I fix my problem by throwing money at it and putting a 2nd proxy at this site and a proxy group?
Proxy settings:
CacheSize=256M
HistoryCacheSize=64M
HistoryIndexCacheSize=32M
ProxyMemoryBufferSize=128M
StartVMwareCollectors=1
# we don't use any vmware at this facility, but I have this as standard on all our proxies
VMwareFrequency=60
VMwarePerfFrequency=60
VMwareCacheSize=16M
ProxyOfflineBuffer=48
StartDiscoverers=3
StartPollers=20
StartSNMPPollers=20
StartPingers=6
StartPreprocessors=6
StartDBSyncers=6
Timeout=3
StartPollersUnreachable=3
MaxConcurrentChecksPerPoller=1000
Graphs below show some atypical activity levels in the last 30 minutes as I've been monkeying with settings (again) to improve performance.
Comment