We just had the PCI riser replaced on a 2850 because the embedded PERC 4 had failed. While reloading the Oracle applications, the DBA reported that 17Gb copy that took 10 minutes on the development server with identical hardware and software (to the extent possible), took over an hour on the prod system with the new board. He showed me "strace cp" results like:
read(3, "\6\242\0\0p\2C\4\27k\324\366\0\0\1\6\241\251\0\0\2\0\0"..., 32768) = 32768 write(4, "\6\242\0\0p\2C\4\27k\324\366\0\0\1\6\241\251\0\0\2\0\0"..., 32768) = 32768 on the fast system and read(3, "\6\242\0\0^H\5\35\340\250\257\270\0\0\1\4\375\265\0\0\2"..., 4096) = 4096 write(4, "\6\242\0\0^H\5\35\340\250\257\270\0\0\1\4\375\265\0\0\2"..., 4096) = 4096 Trying to get clueful about the 4096 vs. 32768 differences and whether they have a bearing on the copy speed differences is presenting me with a steep curve. Can anyone shed any light on this for me? BTW, the only differences I'm aware of on the systems is that the slow system has PERC firmware 5A2D BIOS H433 and the fast one has fw 516A and BIOS H418. RAID adapter and container settings match across systems, as do e2fstune -l settings. No hardware errors showing up in /var/log/messages or dmesg. _______________________________________________ Linux-PowerEdge mailing list Linux-PowerEdge@dell.com https://lists.us.dell.com/mailman/listinfo/linux-poweredge Please read the FAQ at http://lists.us.dell.com/faq
