Hi Neil, The latter half of this update consumes the new md/dev-XXX/recovery_start attribute for checkpointing and resuming resync/rebuild operations. It requires one more kernel patch [1] on top of 2.6.33-rc1 to ensure that mdmon sees the recovery state. One caveat is that we only get checkpoint events when the sync_action == idle, at array shutdown or as a part of 'mdadm --wait-clean --scan'. The other bits are a collection of fixes to address unrelated issues discovered along the way. The result is pushed out to: git://github.com:djbw/mdadm.git master Please have a look. Thanks, Dan [1]: http://marc.info/?l=linux-raid&m=126144478304488&w=2 --- Dan Williams (13): imsm: catch attempt to auto-layout zero-length arrays imsm: honor orom constraints for auto-layout imsm: fix spare promotion imsm: fix thunderdome segfault util: fix devnum2devname for devnum == 0 imsm: cleanup print_imsm_dev() mdmon: cleanup manage_member() leak mdmon: cleanup resync_start Add scaffolding for handling md/dev-XXX/recovery_start Introduce MaxSector Teach sysfs_add_disk() callers to use ->recovery_start versus 'insync' parameter Support external metadata recovery-resume imsm: add support for checkpointing via 'curr_migr_unit' Create.c | 2 Manage.c | 1 bitmap.c | 2 managemon.c | 93 ++++++++---- mdadm.h | 27 +++- mdmon.h | 7 - monitor.c | 23 +-- super-ddf.c | 14 +- super-intel.c | 432 +++++++++++++++++++++++++++++++++++++++++++++++---------- super1.c | 6 - sysfs.c | 15 ++ util.c | 24 +++ 12 files changed, 493 insertions(+), 153 deletions(-) -- 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
