If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to REGISTER before you can post. To start viewing messages, select the forum that you want to visit from the selection below.
The 'Zabbix agent' checks are performed from the zabbix server. It connect to the agent via the IP or DNS name set for that host.
An active check works the other way round. Every few minutes, the agent connects to the server and asks for a list of active checks. If you have say, an active check for CPU load set for every 60 seconds, the agent will then send this value in every 60 seconds.
Active checks are great for hosts behind firewalls as they only require outbound access to the Zabbix server. We use them a lot. They also cut the load on the Zabbix server. The only negative is that it's harder to tell that a host is down if you don't have any non-active checks. You have to set a trigger up to alert you if no data has been received for say, 5 minutes for an item such as the hostname or something else simple.
You can mix and match active and passive checks on the same host if you want. I've seen recommendations in the past, to have just one passive check to allow you to be alerted if the host can't be contacted, and the rest as active checks to reduce load on your server.
The buffer is controlled by two configuration parameters from zabbix_agentd.conf:
BufferSend
BufferSize
We are testing BufferSend and BufferSize.
It seems that the data are send to the server or proxy only when time or size is reached.
So it looks impossible to have a max data preservation in case of server failure and a respectable time and tcp communication to the server in normal case.
Is it possible to have this kind of configuration:
BufferSend=5 # Send every 5 sec in normal mode
BufferSize=100 # Send every 100 items in normal mode
BufferMaxSend=3600 # Buffering 3600 sec in case of server failure
BufferMaxSize=65535 # Buffering 65535 items in case of server failure
Is it possible to have this kind of configuration:
BufferSend=5 # Send every 5 sec in normal mode
BufferSize=100 # Send every 100 items in normal mode
BufferMaxSend=3600 # Buffering 3600 sec in case of server failure
BufferMaxSize=65535 # Buffering 65535 items in case of server failure
We are testing BufferSend and BufferSize.
It seems that the data are send to the server or proxy only when time or size is reached.
So it looks impossible to have a max data preservation in case of server failure and a respectable time and tcp communication to the server in normal case.
Is it possible to have this kind of configuration:
BufferSend=5 # Send every 5 sec in normal mode
BufferSize=100 # Send every 100 items in normal mode
BufferMaxSend=3600 # Buffering 3600 sec in case of server failure
BufferMaxSize=65535 # Buffering 65535 items in case of server failure
Thanks for your help
I would have actually expected that to be the default action for this parameter. I can't see a lot of point buffering all the data when the server is online, it has to be sent eventually so you aren't saving anything, but buffering in case the server is unreachable makes complete sense.
Please do log this as a feature request, it would be useful for me also.
We are testing BufferSend and BufferSize.
It seems that the data are send to the server or proxy only when time or size is reached.
So it looks impossible to have a max data preservation in case of server failure and a respectable time and tcp communication to the server in normal case.
Is it possible to have this kind of configuration:
BufferSend=5 # Send every 5 sec in normal mode
BufferSize=100 # Send every 100 items in normal mode
BufferMaxSend=3600 # Buffering 3600 sec in case of server failure
BufferMaxSize=65535 # Buffering 65535 items in case of server failure
Thanks for your help
please note that, if i understand this correctly, size does not impact when data is sent, but when data is discarded. similarly, time does not impact what is buffered, only when it is sent.
as such, i don't think suggested BufferMaxSize would accomplish anything - you can just set the current buffersize higher.
again, if i understood this correctly, BufferMaxSend would just make agent back off in case server is unreachable - is that really the desired outcome ?
again, if i understood this correctly, BufferMaxSend would just make agent back off in case server is unreachable - is that really the desired outcome ?
Yes, we need parameters for keeping in the agent buffer (postpone data), a certain time or size of data in case of server or proxy unreacheable.
Older data than the BufferMaxSend or BufferMaxSize can be discarded if server is always unreacheable.
hmm, this doesn't seem to match my expectations. apparently there's either a problem, or proper documentation is needed for this - could you please file a new zbx report on the issue tracker ?
Comment