On Tue, Nov 24, 2009 at 10:21:30PM +1100, Ben Finney wrote: > Thanks for the response. > > On 24-Nov-2009, Mark Hindley wrote: > > Can you try the patch at > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=533830#15 > > > > I have had a report that it fixes this bug. > > The patch is already applied AFAICT (my report to this bug specifies > ???apt-cacher??? version 1.6.9), and the behaviour is as I described. So > no, that change doesn't fix the bug.
OK, I have had a look at this again. I can't reproduce it, but I think that on some systems there is a conflict between IO::Select->can_read and alarm(). From the reports I have, I can't see the pattern though. Could you try this patch (against 1.6.9) Mark diff --git a/apt-cacher b/apt-cacher index 8ac5b91..9db5fa0 100755 --- a/apt-cacher +++ b/apt-cacher @@ -1208,7 +1208,6 @@ sub connect_curlm { $active_handles++; while ($active_handles) { - alarm $cfg->{curl_idle_timeout}; # Reset timeout my $active_transfers = $curlm->perform; if ($active_transfers != $active_handles) { if (my ($id,$return_value) = $curlm->info_read) { -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org