Christopher Schultz-2 wrote: > > Rainer, > > On 3/2/2009 11:07 AM, Rainer Jung wrote: >> I think a robust implementation would be something like: >> >> - a new state for temporarily disabling a node, except for sticky >> requests >> >> - a configurable probe request, that the watchdog thread sends, and >> which is answered by the application, the answer containing the >> information whether the node should be temporarily disabled or not > > Just to be clear, you are talking about a new feature that would > temporarily adjust the setting for the "activation" property of a > particular worker in a load-balanced configuration, right? > Yes.
Christopher Schultz-2 wrote: > > This would take a worker from activation=Active to activation=D in > response to some > "configurable probe request" that would be sent by the watchdog thread > periodically. Presumably, the application would eventually respond with > "re-enable this worker". > I would like to have the same behavior of the fail_on_status. When the worker is temporary disabled (activation=D) the connector wait for x seconds. When this time expires it tries to activate the worker, if the status returned is ok the worker will be activated. No threads, no probe. Christopher Schultz-2 wrote: > >> I'm not sure, what the purpose of this is. Assume our session load >> balancing works, so that all three nodes have an equal number of >> sessions, you would soon end up having all your nodes temporarily >> disabled. No? > > I tend to agree with Rainer, here: if you put a cap on your sessions > (per app server) and you are worried about hitting that limit, then you > are likely to hit that limit on all your servers eventually. What, then? > Show a "too many sessions; come back later" page? > Yes, I do. Now it happens when the request reach a tomcat having no more session available. But the LB worker doesn't know and continue to send request to this tomcat. I could use the session method for LB worker, but I have M apache httpd server connected to N tomcat (full-meshed) and I think each LB worker know his session count, but nothing about the other, so the balancing could be "unfair". Isn't it? Thank you very much. Matteo -- View this message in context: http://www.nabble.com/Jk-disable-worker-on-status-tp22289604p22313622.html Sent from the Tomcat - User mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org
