Hi Brain, Thanks for the detailed response.
I build my own kernel with vanilla 2.6.22.19 kernel source + 2.6.22-vanilla patch series and kernel_config files provided by Lustre with no additional patches to lustre or the kernel. I have stopped using the Redhat kernels on our servers as the 1GHz clock used by the Redhat kernels causes a lot of timeouts/errors under heavy I/O. 2.6.22.19 kernel with 250MHz clock is pretty much rock solid on our servers. I did a fresh kernel build with 2.6.22.19 + Lustre 1.8.0.1 patches yesterday on our test machines and it worked fine. When I reverted back to the kernel build with 2.6.22.19 + Lustre 1.8.1 patches it worked as well without any problem. I am confused on why I am not able to reproduce these errors on the test machine while it is consistently reproducible on our production servers. I could possibly try it again on my production server sometime in the next few days. I will also take a close look at the patches and try adding them incrementally. If I understand it correctly, I could apply the patches from Lustre-1.8.0.1 and build the 2.6.22.19 kernel and then build/install Lustre-1.8.1 against that kernel. Thanks Nirmal _______________________________________________ Lustre-discuss mailing list [email protected] http://lists.lustre.org/mailman/listinfo/lustre-discuss
