We ran into this problem also when using a one-tier, multi-point cluster. To
reduce the problem we switched to a two-tier multipoint cluster, turned
BackhandConnectionPools off, and turned off keep alives. I still find that
the problem reappears once any of the servers in first tier servers hits
it's max clients. The only solution i found to stop that was to add more
servers to the second tier.
I would be interested in hearing about other configurations out there and if
others are running into this problem and their solutions.
[mailto:firstname.lastname@example.org]On Behalf Of Dave Lyon
Sent: Friday, May 16, 2003 1:13 PM
Subject: Re: [m_b_users] Interrupted System Call
We found a post about it earlier and disabled connection pooling as he
did and we are no longer receiving the error. We were getting hundreds
of lines of this error every 4 seconds or so on one of the two web
servers. Thanks for the reply.
On Fri, 2003-05-16 at 13:00, Theo E. Schlossnagle wrote: > Dave Lyon wrote:
> > I'm seeing a ton of these errors:
> > [Thu May 15 15:53:36 2003] [warn] (4)Interrupted system call: connect()
> > timed out
> > They only happen on one of our test web servers. Any ideas or
> > suggestions? Thanks.
> That would happen if a machine receiving a request attempts to backhand it
to > another machine that just went down and hasn't yet being marked down yet
by > mod_backhand.
backhand-users mailing list