PURGE delay?

Hello,
I am trying to set up Squid to do something like what HTTP-Accelerator mode
is intended for. However, I want to be able to allow my application to
signal to Squid that a given page has expired. It seems like the PURGE
command is the best way to do this. However, I seem to be having some
performance issues, which is odd, given that I'm currently the only one
using the cache.

My problem is that I issue a:
client -m PURGE <cache-testing-url>

... and get back a:
HTTP/1.0 200 OK
Server: Squid/2.2.STABLE4
Mime-Version: 1.0
Date: Mon, 30 Aug 1999 22:32:07 GMT
Content-Length: 0
X-Cache: MISS from <test-machine>
Proxy-Connection: close

... then for the next minute or so, all of the requests for
<cache-testing-url> give cache misses. More specifically I get pairs like
this in my store.log:
936052930.489 RELEASE 00000017 200 936052994 -1 -1
text/html -1/1
18 GET <cache-testing-url>
936052930.503 SWAPOUT 00000017 200 936052995 -1 -1
text/html -1/1
18 GET <cache-testing-url>

Then after about a minute, it goes back to getting cache hits.

This delay is a pretty big problem for the application I am trying to build.
Any help would be greatly appreciated.

Also, I noticed that if there's an I/O going on for a given object in the
cache, that PURGE requests are queued. Does anybody know if it's therefore
possible for a PURGE request never to get processed (given that there might
always be an I/O request on a popular object)?

Thanks for your help.

later,
graham
Received on Mon Aug 30 1999 - 17:22:40 MDT