[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#500838: linux-image-2.6.24-etchnhalf.1-686-bigmem: MD raid5 array with XFS filesystem hangs (deadlock) after a while under heavy use



Package: linux-image-2.6.24-etchnhalf.1-686-bigmem
Version: 2.6.24-6~etchnhalf.5
Severity: important

File operations hang after a while on a raid5 MD array with XFS filesystem on it. XFS may be irrelevant. Default parameters were used, so /sys/block/md*/md/stripe_cache_size was not changed.
No error messages appear in dmesg or elsewhere, so a deadlock is suspected.

A patch exists: http://www.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.24-rc8/2.6.24-rc8-mm1/broken-out/md-fix-an-occasional-deadlock-in-raid5.patch.
The patch seems to appear (slightly modified) in the 2.6.26 (testing) kernel.
The patch, or the version that appears in 2.6.26 is reported to solve the problem (I can not yet confirm).

Increasing /sys/block/md*/md/stripe_cache_size to 4096 or 8192 is also reported to solve the problem (I can not yet confirm).

The stable 2.6.18 kernel is not affected, though raid5 speed benefits from increasing /sys/block/md*/md/stripe_cache_size (confirmed).

I use lenny/testing with a stable kernel.

-- Package-specific info:

-- System Information:
Debian Release: lenny/sid
  APT prefers testing
  APT policy: (500, 'testing'), (500, 'stable')
Architecture: i386 (i686)

Kernel: Linux 2.6.24-etchnhalf.1-686-bigmem (SMP w/8 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash

Versions of packages linux-image-2.6.24-etchnhalf.1-686-bigmem depends on:
ii  debconf [debconf-2.0]         1.5.22     Debian configuration management sy
ii  initramfs-tools [linux-initra 0.92j      tools for generating an initramfs
ii  module-init-tools             3.4-1      tools for managing Linux kernel mo

Versions of packages linux-image-2.6.24-etchnhalf.1-686-bigmem recommends:
ii  libc6-i686                    2.7-13     GNU C Library: Shared libraries [i

-- debconf information excluded



Reply to: