[Gluster-devel] Lot of Netbsd regressions 'Waiting for the next available executor'

[Raghavendra Talur]

Hi,

I got info from one of the slave machines,

bt from perfused process

#0  0xbb679a77 in _sys___kevent50 () from /usr/lib/libc.so.12
(gdb) info threads
  Id   Target Id         Frame
* 1    process 26515     0xbb679a77 in _sys___kevent50 () from
/usr/lib/libc.so.12


bt from mount process
(gdb) bt
#0  0xbb3bb4b7 in ___lwp_park60 () from /usr/lib/libc.so.12


Manu, this seems to be a bug in libperfuse and not in Gluster.
The machine is nbslave75.cloud.gluster.org. You will have to rerun
quota.t couple of times to hit the bug. The test would hang in line 62(TEST
24).


Thanks,
Raghavendra Talur






On Thu, Dec 24, 2015 at 3:32 PM, Emmanuel Dreyfus <manu at netbsd.org> wrote:

> On Thu, Dec 24, 2015 at 01:44:21AM -0500, Raghavendra Gowdappa wrote:
> > > Seems to be hung. May be a hung syscall? I've tried to kill it, but
> seems
> > > like its not dead. May be patch #12594 is causing some issues on
> netbsd. It
> > > has passed gluster regression.
>
> ps -axl shows PID 1394 (umount) waiting on tstile, which is used for
> spinlocks. No process should sit there for long, unless there is a
> kernel locking problem (which may be a userland locking problem
> thanks to FUSE).
>
> Using crash(8) I can see umount is awaiting for a vnode lock. There
> is certainly something to investigate but I lack time for now. I
> issued a reboot. Please tell me if you can reproduce it.
>
> --
> Emmanuel Dreyfus
> manu at netbsd.org
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20151231/46d34657/attachment.html>