I noticed the vmware polling mechanism creates and breaks-down API logins each time it polls. I noticed this when I saw my VMware hosts' event histories begin to explode with Login/Logout entries after I enabled Zabbix polling. For efficiency's sake and to reduce VMware log growth/rotation, is it possible to pool login sessions across some configurable timeperiod?
Example, VMwareCollector logs in to a host, stores a pid of some sort containing session details, then each subsequent poll uses that same session until: 1) arbitrary configured timeperiod expires 2) a Zabbix-external vmware timeout of some sort kills the session. Either condition would just require the collector re-login next poll.
I'll admit, the concept stemmed from this PRTG forum post:
While VMwareFrequency is helpful in reducing the quantity of API logins logged, we lose granularity in our measurement as well as response time for triggers.
I may just be overlooking something already-existent. I could also probably proxy requests somehow to achieve the goal. Just looking around here in the forums as a first step.
Keep up the great work, Zabbix devs! Happy user here since 1.x.
Example, VMwareCollector logs in to a host, stores a pid of some sort containing session details, then each subsequent poll uses that same session until: 1) arbitrary configured timeperiod expires 2) a Zabbix-external vmware timeout of some sort kills the session. Either condition would just require the collector re-login next poll.
I'll admit, the concept stemmed from this PRTG forum post:
While VMwareFrequency is helpful in reducing the quantity of API logins logged, we lose granularity in our measurement as well as response time for triggers.
I may just be overlooking something already-existent. I could also probably proxy requests somehow to achieve the goal. Just looking around here in the forums as a first step.
Keep up the great work, Zabbix devs! Happy user here since 1.x.