Bug #9185
closed
incorrect Centos 6.5 fastcgi package
Added by Yehuda Sadeh over 9 years ago.
Updated over 9 years ago.
Description
The fastcgi package that is being installed is, or either based off: fcgi-2.4.0-10.el6.x86_64. Not 100% sure that it even has our modifications to it, but in any case, it's based on 2.4.0 and missing the -idle-timeout option. We need to build a newer one that is based of 2.4.7.
- Priority changed from Normal to Urgent
So this problem is with the fcgi package not mod_fastcgi?
fcgi? how does that even enter into it? I thought our work was only with fastcgi?
Is this on teuthology, or customer systems?
(09:51:57 AM) sagehm@newdream.net/montreal: mod_fastcgi-2.4.7-1.ceph.el6.x86_64
(09:52:15 AM) sagehm@newdream.net/montreal: fcgi-2.4.0-10.el6.x86_64
both of these were installed... they are probably misnamed or something
i will reproduce the environment
- Assignee set to Sage Weil
- Status changed from New to In Progress
ok, the idle timeout is working fine.. i can pause the radosgw process (kill -STOP) and curl will block for well over 30s. conversely, if i change idle timeout to 10s, it returns with a 500 after exactly 10s.
but s3tests are still erroring out. there must be a client-side boto timeout?
- Status changed from In Progress to Rejected
this isn't the package, it's boto timeouts.
we can now adjust them up some, but there seems to be a limit of ~3-4 minutes before we get a 'timed out' error anyway.
Also available in: Atom
PDF