zfs discuss - Apr 2010 - ZFS destroy snapshot stall writes? (snv_130).

I''m using EON 0.60.0 based on svn_130. I populated a 13.5 TB mirrored
ZFS volume with serveral large files (dd from /dev/zero, ~100GB) and created a
handful of snapshots. NO DEDUPE.

Having deleted some of the files, the snapshots grew considerably:

tank/tst at 140410075500  1.12T      -  8.17T  -
tank/tst at 140410083501  7.87M      -  7.57T  -
tank/tst at 140410093502   201G      -  8.35T  -
tank/tst at 140410105501   257G      -  5.92T  -

Now, destroying some of the snapshots (it''s size doesn''t seem
to matter) under I/O load stalls all writes to the pool for up to 1 minute.

The filesystem is NFS-exported so write stalls obviously create considerable
havoc.

Is this an known issue? Has anyone seen this behavior?

Thank  you.
-- 
This message posted from opensolaris.org