]> git-server-git.apps.pok.os.sepia.ceph.com Git - ceph.git/commit
ECBackend::continue_recovery_op: handle a source shard going down
authorSamuel Just <sam.just@inktank.com>
Thu, 24 Apr 2014 19:48:44 +0000 (12:48 -0700)
committerSage Weil <sage@inktank.com>
Wed, 30 Apr 2014 22:08:32 +0000 (15:08 -0700)
commitd9106ce5e4437ab02279c4d7dfa23ce0d69c59df
tree6882efea53de51f411e03590295b1505c6524698
parent87195d5f69ddd8060035882ccc760d25964773a0
ECBackend::continue_recovery_op: handle a source shard going down

get_min_avail_to_read_shards might return an error if there are
no longer enough sources to reconstruct the missing shards.
This is possible if osds went down while we were writing the
previous chunk -- we already notice in check_recovery_sources
if a source goes down during a read.

Fixes: #8161
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 1885792c517670086332a8bab237c58558ee6dda)
src/osd/ECBackend.cc