$ua->Timeout(n) doesnt
jb@dslreports.com
Mon, 11 Oct 1999 13:43:24 -0400
Hi.,
I've done my due diligence with deja, www.google.com/linux and this
list archive, and I cant find an answer.
Redhat 6.0 (two different boxes, one with a freshly recompiled newer
kernel), exhibit the above mentioned problem.
The trace shows the alarm(n), the connect(), the alarm goes off, connect
returns ERESTARTSYS
sigreturn(), connect() starts up again, to time out ages later at
the tcp stack timeout value i suppose.
If you point $ua->request at a downed IP, or a malfunctioning web server,
your process is hung up on that for minutes.
What can I do to get around this.. its driving me crazy, i have to write
hack wrappers for some scripts that cannot be allowed to hang like this..
I am kind of tempted to change the tcp config to reduce connect timeouts
to 10 seconds, but I'm afraid of unforseen consequences as that would
be applied to everything?
I've seen others report variants of this problem in other places
on the web, but no answers that I could find.
thanks for any help
-Justin