Solaris and siteminder nasty interaction?

[Bill Stoddard <bi...@wstoddard.com>]

We have several bug reports open on Apache 1.3 on Solaris reporting the
following error in the error log:


fcntl: F_SETLKW: Error getting accept lock, exiting!  Perhaps you need to use
the LockFile directive to place your lock file on a local disk!
[Wed Oct 18 14:03:31 2000] [notice] child pid 19796 exit signal Segmentation
Fault (11)
[Wed Oct 18 14:07:59 2000] [emerg] (9)Bad file number: fcntl: F_SETLKW: Error
getting accept lock, exiting!  Perhaps you need to use the LockFile directive
to place your lock file on a local disk!


In each case, it appears the the the lockfile directive is coded and the lock
file is on a local file system (not on a network mounted file system). I see
two cases where someone reported that they were also running Netegrity
Siteminder.  This strikes me as an odd coincidence.  Furthermore, one of the
bug reports states...

<quote>
This problem is caused by another application that can cause time
delays.  For example, a security agent running at the server.

Resolution is to compile Apache with the following flags.
CFLAGS -DUSE_SYSVSEM_SERIALIZED_ACCEPT
</quote>

I can appreciate compiling with USE_SYSVSEM_SERIALIZED_ACCEPT, but I
completely do not get the first statement. "time delays"? And how could this
possibly affect fcntl lock failure?

I want to know if -the- solution is using a different accept lock or is is
there another solution. Can't make that determination until we understand wtf
this failure is happening anyway.

Any ideas?

Thanks,
Bill


________________________________________________
Bill Stoddard stoddard@raleigh.ibm.com

Re: Solaris and siteminder nasty interaction?

[Jeff Trawick <tr...@bellsouth.net>]

"Bill Stoddard" <bi...@wstoddard.com> writes:

> We have several bug reports open on Apache 1.3 on Solaris reporting the
> following error in the error log:
> 
> 
> fcntl: F_SETLKW: Error getting accept lock, exiting!  Perhaps you need to use
> the LockFile directive to place your lock file on a local disk!
> [Wed Oct 18 14:03:31 2000] [notice] child pid 19796 exit signal Segmentation
> Fault (11)
> [Wed Oct 18 14:07:59 2000] [emerg] (9)Bad file number: fcntl: F_SETLKW: Error
> getting accept lock, exiting!  Perhaps you need to use the LockFile directive
> to place your lock file on a local disk!

Hmmm... look at that strerror() message: "Bad file number"

Is it possible that some code loaded into the server is inadvertently
closing the file descriptor for the lock file?

-- 
Jeff Trawick | trawick@ibm.net | PGP public key at web site:
     http://www.geocities.com/SiliconValley/Park/9289/
          Born in Roswell... married an alien...

Any objections setting default to USE_SYSVSEM_SERIALIZED_ACCEPT on Solaris?

[Bill Stoddard <bi...@wstoddard.com>]

For 1.3.15?

Bill