On Sat, 6 Jan 2001, Adrian Chadd wrote:
> Note again that this is only happening on the modio branch for now.
> If noone objects I'll port over the patches to squid-HEAD to tidy
> things up a little.
I am afraid your change will effectively disable Cache Digests. It is
probably prohibitively expensive to generate digests from scratch if
every object's meta-data has to be read from disk.
I see two options:
- Keep all your changes to your branch only. Folks who
want to test modio branch will be warned that digests should
be disabled and ICP might be too slow.
- Change the way Cache Digests are computed/maintained.
In short, one would have to maintain a digest-like
memory-resident hash that will be used to answer
ICP queries and build digests if needed. The hash
cannot be the same as current digests because we
probably need to support delete operations.
Your changes can be brought in to HEAD after the above
is done and tested.
Overall, I would think that such a drastic change as making storeGet()
blocking should be well tested (at least HTTP/ICP/Digest-performance
wise) in isolation before it is admitted into mainstream code.
$0.02,
Alex.
Received on Sat Jan 06 2001 - 10:58:54 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:13:11 MST