apache-api/1035: Servers not starting correctly. Looks like file locking problem.

[Justin Ashworth <ju...@ashworth.org>]

>Number:         1035
>Category:       apache-api
>Synopsis:       Servers not starting correctly. Looks like file locking problem.
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    apache (Apache HTTP Project)
>State:          open
>Class:          sw-bug
>Submitter-Id:   apache
>Arrival-Date:   Thu Aug 21 15:10:01 1997
>Originator:     justin@ashworth.org
>Organization:
apache
>Release:        1.2.3
>Environment:
SunOS www 5.5.1 Generic_103640-08 sun4u sparc SUNW,Ultra-1
gcc -O2
Proxy patch although it probably doesn't matter in my case.
>Description:
>From error_log:

......
[Thu Aug 21 14:48:37 1997] fcntl: F_SETLKW: No record locks available
[Thu Aug 21 14:48:37 1997] - Error getting accept lock. Exiting!
[Thu Aug 21 14:48:37 1997] fcntl: F_SETLKW: No record locks available
[Thu Aug 21 14:48:37 1997] - Error getting accept lock. Exiting!
[Thu Aug 21 14:48:37 1997] fcntl: F_SETLKW: No record locks available
[Thu Aug 21 14:48:37 1997] - Error getting accept lock. Exiting!
[Thu Aug 21 14:48:38 1997] fcntl: F_SETLKW: No record locks available
[Thu Aug 21 14:48:38 1997] - Error getting accept lock. Exiting!
[Thu Aug 21 14:48:38 1997] fcntl: F_SETLKW: No record locks available
[Thu Aug 21 14:48:38 1997] - Error getting accept lock. Exiting!
.......

This caused the browser (Netscape Communicator 4.02) to just sit and wait with the "waiting for reply...." status.
Only the root process started on the server...no child processes were spawned.
Not all of our virtuals (the older-style virtuals, not HTTP/1.1) had this problem although all were using the same copy of the httpd.
Some of the virtual servers had links to NFS-mounted drives.
Server load increased nearly 10-fold as well.
>How-To-Repeat:
Try NFS-mounting a drive and having HTML documents exist on the other side of that mount. See if that will do it.
>Fix:
Nope, but I would love to know when you hear of one. Had this problem with 1.2.1 as well but due to the amount of bugs with that we just reverted back to 1.2.0. We are running 1.2.0 right now
>Audit-Trail:
>Unformatted: