[squid-users] Painless rebuilds question

Hello
 I have been testing something to try to let an accellerator cache rebuild,
without impacting the origin server when first starting. I am having it build it
its cache store by using the round-robin parent setup below. This makes the
cache server (#3 of 4) get its contents off the other 3 cache servers. Then,
after it has been on for awhile, and built up a sizeable cache, I switch it back
to normal mode.

 Sometimes this works very well. However many times, when I cut it back to
normal setup, it really slams the origin server hard for at least an hour or
two, then slowly starts to request less heavely. Since I have let its cache
build up, I do not understand why it is slaming the origin server so much. Is
there something in the data that it gets via the other caching servers that
would make it need to test against the origin server again? Is my idea flawed?

 Any suggestions most appreciated.

## Normal
#cache_peer 1.1.1.1 parent 80 0 no-query default
#cache_peer 1-1.server.com sibling 80 3130
#cache_peer 1-2.server.com sibling 80 3130
#cache_peer 1-4.server.com sibling 80 3130

## To let it rebuild
cache_peer 1-1.server.com parent 80 3130 round-robin
cache_peer 1-2.server.com parent 80 3130 round-robin
cache_peer 1-4.server.com parent 80 3130 round-robin

 Thanks!

   Nicole
Received on Tue Aug 21 2007 - 16:04:37 MDT