Forum Discussion
NimbostratusRetry downed pool member
We are trying to implement some passive monitoring for our sites/pools by the use of the inband health monitor and the use of an irule to look for errors in the http response. Reference Article Overview of passive health monitoring
I am able to "no-brain" copy paste the exampled iRule thats recommended and change the text to look for something like service unavailable. Ideally I'd want to change that to a data group or something to look for other lines of text as well. But for now I can stop an app pool on a site and cause the member to go down because a 503 is being generated. Is there a way I can have that pool member retested and if it doesn't give an error enable the pool member again? Or if it's still down keep it down and pass the traffic to the other up members?
3 Replies
IainThomson85_1Cumulonimbus
Does it need to be passive monitoring?
If you're looking at the HTTP response, would it not be easier to look for a "Good" response from the server ?
Apologies if this was already an option that was considered and discounted.
- IainThomson85_1
Might be worth adding some RegEx in the receive disable for 50X codes like the below.
^HTTP.1.[0|1]\s([5][0-9])
Ed_SummersThe inband monitor does have a retry-time after which it will allow the pool member to be tested again. However you may be looking for an active monitor in conjunction with inband. Active monitors have a property called "up-interval" that allows setting a different polling interval for pool members that are UP vice DOWN (downed pool members would be polled at the rate specified by the 'interval' property). By default up-interval is the same as interval, so the polling interval is the same regardless if the server is up or down.
I cannot tell if you can disable the up-interval - I'd have to lab it later or you can test on a test pool. You could set up-interval to something long so the amount of traffic generated would be negligible. Make sure you configure that monitor with the appropriate checks.
