On Tue, 2003-03-11 at 10:40, Andres Kroonmaa wrote: > On 11 Mar 2003, at 15:55, Henrik Nordstrom <[EMAIL PROTECTED]> wrote: > > > > The worse problem, however, is once this probelm occurs (origing servers > > > "partially" down) squid refuses to serve some objects from cache. The > > > objects are in the cache and fresh, but squid does an IMS to the origin > > > for each request for these objects. > > > > Then you need to dig into the refresh logics of Squid and figure out why > > the IMS replies received does not make Squid consider the object fresh > > for future requests if it has failed to revalidate the object > > previously.. > > Not sure if this could be valid reason, but can it be that when squid > goes for origin check, url is put into "private" state, invalidating > local copy until origin reply allows it to become cached again. And if > origin reply never comes, all new requests for same url have to go to > origin?
This sounds plausible, though it seems as though the origin connection 'should' eventually time out, or does the half closed connections setting somehow interfere with that? > > > Sounds odd that a reconfigure makes Squid snap out of the condition. I > > can understand if restart helps, but not a reconfigure.. > > All connections forcibly closed? I will have to examine this, but I don't think that is it. -- David Nicklay Location: CNN Center - SE0811A Office: 404-827-2698 Cell: 404-545-6218
