]> git-server-git.apps.pok.os.sepia.ceph.com Git - ceph.git/commit
mds: update MDSRank::cluster_degraded before handling mds failure 23169/head
authorYan, Zheng <zyan@redhat.com>
Thu, 2 Aug 2018 02:05:12 +0000 (10:05 +0800)
committerYan, Zheng <zyan@redhat.com>
Fri, 3 Aug 2018 01:14:14 +0000 (09:14 +0800)
commit9b4e8cb03b76cb113e3e07836de380d1c7d2dfcd
treeab4fa4fad701e184c109ec808a3a58c2b337e7cc
parentb0122e2e5b50ba570d4c93982c07a1712a4d5106
mds: update MDSRank::cluster_degraded before handling mds failure

Migrator checks MDSRank::is_cluster_degraded() to decide if it needs to
send message to other mds. MDSRank::cluster_degraded should get updated
before calling Migrator::handle_mds_failure_or_stop().

Introduced by commit 7de9da4a "mds: handle discontinuous mdsmap"

Signed-off-by: "Yan, Zheng" <zyan@redhat.com>
(cherry picked from commit 93458c7306c3822175ed3c048f0937af8f02c341)
src/mds/MDSRank.cc