apache giving errors on start.

[nikhil]

i have a redhat 9 box with kernels 2.4.20-8 and 2.6.0
and i have Apache/2.0.45 running.....the problem is while booting apache starts up correctly....but then if i stop it and start it again i get the following in the error logs:
[emerg] (3Function not implemented: Couldn't create accept lock

this happens with both the kernels....anyone experiencing similar issues?
i googled around a bit and came across posts which faced similar problems and claim they have been resolved but I unfortunately am still experiencing it!!!

[nikhil]

if i do a normal ./apachectl start instead of startssl i get the error
[Fri Jan 23 12:01:44 2004] [error] (3Function not implemented: Cannot create SSLMutex Configuration Failed

[Dolda2000]

That's really weird... have you messed up your futexes or linuxthreads or something?

[nikhil]

hmmm...wat are futexes?
i hvnt messed up anything...its jus that i noticed this behaviour today...tried to restart apache from office and it jus wud die...googling led me to some redhat posts which had filed this as a bug for httpd versions 2.0.40 and below i guess....but i hv 2.0.45 so it should not be a problem....i dont know why its cribbing...
is it trying to create a lock file or somethg?

[Dolda2000]

I'm not sure how that mutex works... Can you try to start apache in a strace session, like this:

strace -o httpd.strace httpd -X

Then, grep the httpd.strace file for any syscall that returns ENOSYS and post which one it is, preferrably with some lines around it.

[nikhil]

i grep'd the httpd.strace file but did not find any lines with ENOSYS but found this at the end of the file:

open("/dev/zero", O_RDWR) = 10
old_mmap(NULL, 24, PROT_READ|PROT_WRITE, MAP_SHARED, 10, 0) = 0x4043e000
close(10) = 0
gettimeofday({1075229237, 691778}, NULL) = 0
open("/etc/localtime", O_RDONLY) = 10
fstat64(10, {st_mode=S_IFREG|0644, st_size=1017, ...}) = 0
old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x4043f000
read(10, "TZif\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\4\0\0\ 0\4\0"..., 4096) = 1017
close(10) = 0
munmap(0x4043f000, 4096) = 0
write(8, "[Tue Jan 27 10:47:17 2004] [erro"..., 8 = 88
write(2, "Configuration Failed\n\n", 22) = 22
close(9) = 0
close( = 0

[Dolda2000]

This is troubling... I would look at the source at apache to try and solve this, but I only have the source for apache 1.3, which works very differently. Can someone solve this, or does anyone at least know of any CVS repository for apache 2 with a HTTP interface that I can look at?

[nikhil]

the problem is I did a search for this bug and it took me to official redhat sites where it said this bug had already been filed and resolved but does not mention anywhere as to how it was resolved in the first place!!!!
everytime i reboot the machine apache comes up successfully, its just if I do a manual stop/start it dies on me....i'll give it a shot with apache1.3 and see how it goes.