On Tue, Feb 14, 2006 at 04:36:00PM -0500, Andrew Pochinsky wrote: > Hi, > I see some strange behavior on a SLES9 client v 1.3.2 running on > power5. After some time the system log (dmesg) is full of messages like > the following: > > Badness in page_remove_rmap at mm/objrmap.c:383 > Call Trace: > [c000000072227950] [c0000000000b1a70] .unmap_vmas+0x67c/0x914 > (unreliable) > [c000000072227a80] [c0000000000b4114] .exit_mmap+0xbc/0x2c4 > [c000000072227b20] [c000000000060200] .mmput+0xb4/0x100 > [c000000072227bb0] [c000000000066868] .do_exit+0x378/0x1118 > [c000000072227c80] [c000000000067654] .do_group_exit+0x4c/0xf0 > [c000000072227d10] [c000000000010464] .ret_from_syscall_1+0x0/0xa4 > --- Exception: c00 at 0xfeb6718 > LR = 0xff5664c > > The system show very poor PVFS2 performance and pvfs2-client-core takes > a lot of cycles. The file system is under moderate load at this time. I > don't know if the dmesg is a cause or the result of pvfs2 degradation. > Has anybody seen such a behavior?
Hi Andrew kernel comments say 'do_group_exit' is called by fatal signals or the exit_group system call. Can you check pvfs2-client.log? If something is causing pvfs2-client-core to die and get restarted often, maybe it will show up in the log file? ==rob -- Rob Latham Mathematics and Computer Science Division A215 0178 EA2D B059 8CDF Argonne National Labs, IL USA B29D F333 664A 4280 315B _______________________________________________ Pvfs2-users mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
