On 02/03/2014 03:22 PM, Alex Rousskov wrote: > On 02/03/2014 01:55 PM, Amos Jeffries wrote: >> FWIW; this seems to be the same issue is under discussion in squid-users >> thread "rock store: a bug or ...?". Cc'ing Henrik Lidström and Nikolai >> Gorchilov. >> >> Henrik, Nokolai: if you could followup to squid-dev in future about this >> please.
>> On 2014-02-04 05:25, Kinkie wrote: >>> Hi, >>> it appears that the large rock merge has introduced a regression, >>> probably related to the handling of keepalive and/or about finishing >>> to send an object to a client. >>> >>> I have confirmed that the issue is NOT present in revno 13200, but it >>> is present in r13203 (r13201 and r13202 do not compile). > FWIW, if nobody volunteers earlier, I should be able to look at this on > Wednesday. FYI: Kinkie and I fixed the bug we could reproduce. See trunk r13257. The bug was a side-effect of the following Collapsed Forwarding change: > revno: 12501.1.50 > committer: Alex Rousskov <rouss...@measurement-factory.com> > branch nick: collapsed-fwd > timestamp: Tue 2013-06-25 09:39:10 -0600 > message: > Mark client streams that sent everything as STREAM_COMPLETE. > > The old code used STREAM_UNPLANNED_COMPLETE if the completed stream was > associated with a non-persistent connection, which did not make sense to me > and, IIRC, led to store entry aborts even though the entries were not > damaged > in any way. > > This change may expose other subtle bugs, but none are known at this time. > > See also: > http://www.squid-cache.org/mail-archive/squid-dev/200702/0017.html > http://www.squid-cache.org/mail-archive/squid-dev/201102/0210.html Cheers, Alex.