I'm curious if others have tuned the client timeout settings and have any recommendations.
We make pretty heavy use of Memcached in a web app and for some class of data we have a poor-man's HA setup where we have two pools of client connections (write to both, read from first that doesn't error). Each pool has *only one* Memcached server. If a server is down the failed connection is very fast and the "fail over" works with little added latency. But we had a switch fail in front of one of the Memcached servers and for some reason the result was timeouts instead of a fast failed connection. We use libmemcached and have the MEMCACHED_BEHAVIOR_POLL_TIMEOUT set to 4 seconds. The web code didn't have a circuit breaker so you can imagine how well that went. Short of the whacky setup, just wondering what client timeout are commonly used. -- Bill Moseley mose...@hank.org -- --- You received this message because you are subscribed to the Google Groups "memcached" group. To unsubscribe from this group and stop receiving emails from it, send an email to memcached+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
