Bug#556610: incremental checkarray every night instead of a full monthly

[Sergey B Kirpichev]

Hello,

Attached new patchset.

1) Run checkarray for 1h every day, doing 1/CHECK_SPLIT of job.
2) CHECK_SPLIT tunable via /etc/default/mdadm or by --split option
3) Introduced --interrupt and --cancel options (last one --interrupt
   current arrays check *and* reset sync_min/sync_max to default).

Remaining issue: multiple arrays, sharing common physical disks.  We
can set /sys/block/mdX/md/sync_force_parallel to 1 temporary while
doing checks.  It's easy doable, but may introduce some problems.
Another solution (looks ugly): pooling
/sys/block/mdX/md/sync_completed *and* cancel check when
sync_completed equal to sync_max.

PS: Looks like there is no chance to enter any solution for this
whishlist item to Wheezy.  But it would be really nice to see anything
from the maintainers, related to suggested above implementations.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 556610.patch
Type: text/x-diff
Size: 6631 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-mdadm-devel/attachments/20120614/0d3fc80a/attachment.patch>