Forum Discussion
Aantat
Cirrus
CirrusLosing Requests because of health monitoring
Hi team. I'm facing issue with losing requests because of health monitoring with a simple http application. When we have a unavailable pool member and after few seconds or minutes it becomes availabl...
Aantat I believe the issue you are experiencing is when your pool member is transitioning from green (healthy) to red (failed). These connections will always drop because the F5 still thinks the device is alive because the health monitor hasn't failed yet but is in the process of failing and I do not believe you can get around this at all other than the connections dropping because the F5 doesn't transition a client from one pool member to another until the health monitor fails completely.
Hooni_LCirrus
CirrusHi, I think below option will help.
first option is 'time until up' in monitor.
second option is 'manual Resume' in monitor
ashrafdawoodNimbostratus
Hello Aantat
please make sure that your service is automatically going up when your server become up , this is setting on service tab on server , also there is seting to make service up when sytem become up
AantatHi ashrafdawood
I'm losing requests not when node become up, but when it goes down...
- Aantat
Hi Hooni_L.
I'm facing problem not when node becames available. I'm facing it when node is going down. I confgirud "Action on service down" to Reselect, but I'm still losing 2-3 requests which is critical. My goal is IF one of the nodes is going down - requests goes to another node without losing that request. I hope I made it clear
- Paulius
MVP
Aantat I believe the issue you are experiencing is when your pool member is transitioning from green (healthy) to red (failed). These connections will always drop because the F5 still thinks the device is alive because the health monitor hasn't failed yet but is in the process of failing and I do not believe you can get around this at all other than the connections dropping because the F5 doesn't transition a client from one pool member to another until the health monitor fails completely.