I have an AMD K7 machine with an 8way RADI5 array on a Marvell based
K7 is pretty wimpy by today's standards - really the K8 was AMD's first solid, server-quality platform. in addition, you appear to have it under pretty high memory pressure.
controller (sata_mv). When data is coming in over the Gigabit nic and is written to disk, messages like the one below intermittently shows up on the console. Eventually the machine freezes solid (hardware reset required).
freezing solid was somewhat common in the K7 era, hence the comment about server-quality.
kernel: md0_raid5: page allocation failure. order:0, mode:0x20
this is a straightforward memory shortage; you can probably avoid it by increasing min_free_kbytes. for this kind of server, I also like to encourage the kernel to write file IO back to disk more promptly
than default: # delay writing dirty blocks hoping to collect further writes (default 30s) vm.dirty_expire_centisecs = 1000 # try writing back every 1s (default 500=5s) vm.dirty_writeback_centisecs = 100 (/etc/sysctl.conf or other mechanism.) (the default settings tend to stutter under high write load, since the kernel will accumulate 30 seconds of dirty pages (gigabytes), and then frantically try to write them out...)
pdflush: page allocation failure. order:0, mode:0x20
right. nothing buggy here, just poorly tuned. need to figure out how to avoid getting this low on memory. (you don't have swap disabled, do you?)
[<c0184331>] reiserfs_sync_fs+0x2c/0x46
hmm. reiserfs doesn't have a great rep for server-type stability either. - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
