We're looking at deploying Zabbix in our 1,000+ server environment and want to put our proxies behind a load balancer. This way should we need additional proxies, I can just add them to the load balancer and not have to tell each client that we've added some. Has anyone done this? Not wanting to re-invent the wheel if it's already been done...
-
-
solution..
After discussions amongst IRC and the Zabbix training course, it doesn't look like we can load balance per se, but there is somewhat of a solution for us.
Our plan is to use 100% passive proxies and agent checks. While this is a little limiting(no log monitoring for example) we're still in a good place because now if we for example lose a proxy at a site, we can just tell Zabbix to start routing through the other proxy.
The biggest thing I see us losing by using totally passive elements is log monitoring. However we're also a heavy splunk shop and ship a ton of stuff into splunk already. So what I've done is write a script that pulls log data from splunk and ships it back to Zabbix using zabbix_sender, which I can then trigger off of log entries I'm looking for. We also use powerbroker alot, so if anything comes up that I need to trap that I'm missing because of a lack of passive checks, I can likely use powerbroker and an external command to get it. -
It would great to have ability to assign a host to a group of proxies for both HA and load balancing perhaps also redundancy. It's not a trivial piece of work but could be included into 2.2 roadmap if financially supported. Let me know if you are interested.Comment
-
Groups of Proxies for HA will be a masterpiece for Zabbix and solved one of our problem
Comment
-
Proxy groups would be a great benefit in our environment as well.Comment
Comment