Proxy has high data sender processes

[mellis]

Good Afternoon

We have a proxy that always has high sender processes over 75%. We are running a 4.4.10 server with a 4.4.6 proxy. This proxy is in a AWS cloud setup with 4cpu's, 8GB ram, disk space is not an issue. We have about 20 other proxies that are in physical data centers all in the states. We have work loads that go from as little as 15 host to 700 host in these data centers. We have been able to tune all the other proxies to handle the load, but this one has been a problem from day one.

The one thing i did notice is it looks like agent items are up to 8 hours old, but on the proxy the internal items seem to be current, i change the agent.ping to and agent(active) item and it seems to clear up and become current.

from the proxy screen
9 host, 546 items, .78 vps

proxy config:
ProxyOfflineBuffer=4
# HeartbeatFrequency=60
# ConfigFrequency=3600
DataSenderFrequency=5
StartPollers=48
StartTrappers=24
StartPingers=6
StartDiscoverers=6
StartHTTPPollers=8
CacheSize=128M
StartDBSyncers=8
HistoryCacheSize=256M
HistoryIndexCacheSize=32M


We are using MySQL as the database
innodb_buffer_pool_size = 1024M
innodb_file_per_table=1
join_buffer_size = 256M
sort_buffer_size = 4M
read_rnd_buffer_size = 4M.

Attached is a word doc that have more details

I am wondering if the agent data is being delayed by the AWS for some reason?

[ingus.vilnis]

Hi,

I'd recommend first to default these values as you have set too many for a proxy which handles virtually no load at all.

DataSenderFrequency=5
StartPollers=48
StartTrappers=24
StartPingers=6
StartDiscoverers=6
StartHTTPPollers=8
CacheSize=128M
StartDBSyncers=8
HistoryCacheSize=256M
HistoryIndexCacheSize=32M

Next step in troubleshooting the connection issues is to see if the proxy can successfully reach Zabbix server.

telnet your.zabbix-server.com 10051

Check if the connection is made successfully and check how much time does it take.

One thing that I had myself recently when proxy data sender went busy - DNS settings of that server pointed to one decommissioned DNS server causing a delay in all the network communications between that proxy as it tried to resolve to non existent addresses.

[mellis]

Well i found the issue, turns out a systems administrator installed trend micro cloud one on the systems, this eat the processors up scanning all network and disk io.
once i stop these services the system started work and the data sender dropped below 10% usage.

I was then able to lower my settings and all is happy.