Hi Derek
Sorry i may have wasted your time > You don't say which particular kernel you are running. You're running > the 'smp' kernel? I am testing it on a mix of UP and SMP RH7.3 boxes with redhat kernel rpms 2.4.18-3 and 2.4.18-4 for both type of machine > >>iget4 smp_e0311e23 - 2.4.18-3-i686.mp >> > sock_create=smp_083c8c04 >>iget4 smp_e0311e23 - 2.4.18-3 >> > sock_create=smp_083c8c04 > > > Hmm, I'm not sure what "2.4.18-3" is from. I also notice that you > don't have a .bm kernel. Very odd... I'm going to have to take a > look at this. Unfortunately I don't have a very good algorithm to > decide which kernels to build. The fact that iget4 and sock_create > are the same for both kernels is also a problem, but I'd be interested > to know why it's not working. > > Very odd indeed. > We are creating our own non-transarc rpms using your src.rpms as a base. We must have made a mistake in the modifications to the spec file because 2.4.18-3 not present in your rpms. Interestingly the md5sums of libafs-2.4.18-3.o and libafs-2.4.18-3-i686.mp.o are different 8140ce8849d461546bce9f61bc311a91 libafs-2.4.18-3-athlon.mp.o dc34f2d32fd9bdd7ecf6cc5700c0b025 libafs-2.4.18-3-athlon.o cf693d82d691d2a9d1a891ee0df7faf4 libafs-2.4.18-3-i386.mp.o 0b2b934807b3e903e403bd9a70fdb114 libafs-2.4.18-3-i386.o 30ea0bba3ca2a04b995fa554b49bea97 libafs-2.4.18-3-i586.mp.o 4ad4efbd623050c1ed8668bd4b5573c9 libafs-2.4.18-3-i586.o ae180b7cd48eec9fa5c35c5a8e737dee libafs-2.4.18-3-i686.ep.o 60bc4356570d301abcaa2200ed6ada0d libafs-2.4.18-3-i686.mp.o 7eb59db048a9a861b99126ea6dcd90b3 libafs-2.4.18-3-i686.o a48d826490ed486e9be7616aa2d19d42 libafs-2.4.18-3.o >>This causes all sorts of problems when the libafs-2.4.18-3.o kernel >>module tries to load. I fixed it by deleting the "iget4 smp_e0311e23 > > > Ok, what sorts of problems? > The module would load but it killed the networking. We had to reboot the machine with afs off > >>2.4.18-3" line. But is there a better way? > > > Well, there shouldn't be a "2.4.18-3" line; I suspect that a patch I > asked for 1.2.4 didn't make it into the 1.2.5 release -- I'll have to > go take a look. > Our screw up I am afraid > >>Also on one machine there is no iget4 lines in /proc/ksyms > > > Did you compile your own kernel on this machine? It sounds like you > recompiled a kernel without kernel symbols. > > No it is a redhat kernel update (kernel-2.4.18-4.i686.rpm) for problems with ext3 in 2.4-18-3. Apparently there is a bug (http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=64838) in redhat kernel-2.4.18-4 which truncates the cat/grep output of /proc/ksyms (dd gets a full file output) and iget was one of the things lost. Fixed in 2.4.18-5 rpms. >>Again a manual edit of the init.d script fixes it. > > > How did you edit the init script to fix this? > I just set LIBAFS=libafs-2.4.18-4-i686.o Cheers Matt _______________________________________________ OpenAFS-info mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-info
