On 25/03/2013 05:55, sugusbs wrote:
1. Sibling A receives 1st request for R and responds with tcp_miss and
FIRST_UP_PARENT.
2. Sibling B receives 2nd request for R and responds with sibling_hit
due to sibling A.
3. After R object gets expired, sibling A receives 3rd request for R and
responds with tcp_miss and FIRST_UP_PARENT.
4. Sibling B receives 4th request for R and should respond with
sibling_hit but tcp_miss and FIRST_UP_PARENT.
The object R has been refreshed in step 3, why does sibling B responds with
FIRST_UP_PARENT in step 4?
I have the same question, using squid 3.1 It seems that whenever an
objects exists in the cache but is stale, other siblings are not checked
for a fresher object but only the origin server.
Is it possible to force a check of the object via a sibling first ?
if the object doesn't exist siblings are queried correctly, it's only
when an object exists but is stale.
Michael
