One more attempt: stuck processes
Gary Mills
mills at cc.umanitoba.ca
Fri Nov 16 09:00:07 EST 2007
On Fri, Nov 16, 2007 at 01:54:24PM +0100, Alain Spineux wrote:
> On Nov 16, 2007 12:36 PM, Sebastian Hagedorn <Hagedorn at uni-koeln.de> wrote:
> > --On 16. November 2007 11:27:09 +0100 Sebastian Hagedorn
> > <hagedorn at uni-koeln.de> wrote:
> >
> > 1. In the absence of the SO_KEEPALIVE option it is entirely possible that a
> > TCP connection remains ESTABLISHED even when the other side has gone.
>
> I said that socket should timeout, but this is true only when the
> protocol (TCP here)
> require a response (usualy AK here) or at connection establishement.
> On the contrary
> it should stay open indefinitely util something happens. Router doing
> NAT can drop
> a too old connection, because it has to maintains a NAT table and make some
> cleanup time to time, this where "KEEPALIVE" become usefull.
>
> > This may not be a solution to this particular problem, but it made me
> > wonder why Cyrus does *not* use SO_KEEPALIVE. Is there a downside to it?
>
> Cyrus has already a built-in time out, it seems a lite conflicting to actively
> maintains the connection until it drop it itself !
> This is the works of the client to actively maintains the connection,
> if it want it !
This timeout doesn't work in some cases. We have lots of POP sessions
that never terminate. About 30 out of 40 are in that state now.
Here's an example:
cyrus 13075 708 0 Oct 14 ? 0:05 pop3d -s
cyrus 20023 708 0 Oct 29 ? 0:00 pop3d
cyrus 24560 708 1 07:38:03 ? 0:03 pop3d
cyrus 631 708 0 Oct 03 ? 0:10 pop3d -s
cyrus 6786 708 0 Oct 20 ? 0:00 pop3d -s
cyrus 29777 708 0 07:45:03 ? 0:00 pop3d
cyrus 19175 708 0 Oct 04 ? 0:04 pop3d -s
One I just checked is stuck in a read():
# truss -p 19175
read(0, 0x002316F0, 5) (sleeping...)
^?# pfiles 19175
19175: pop3d -s
Current rlimit: 256 file descriptors
0: S_IFSOCK mode:0666 dev:271,0 ino:25813 uid:0 gid:0 size:0
O_RDWR
sockname: AF_INET 130.179.16.23 port: 995
peername: AF_INET 130.179.188.184 port: 51771
--
-Gary Mills- -Unix Support- -U of M Academic Computing and Networking-
More information about the Info-cyrus
mailing list