From: Xiubo Li <xiubli@xxxxxxxxxx> commit e7e607bd00481745550389a29ecabe33e13d67cf upstream. Flushing the dirty buffer may take a long time if the cluster is overloaded or if there is network issue. So we should ping the MDSs periodically to keep alive, else the MDS will blocklist the kclient. Cc: stable@xxxxxxxxxxxxxxx Link: https://tracker.ceph.com/issues/61843 Signed-off-by: Xiubo Li <xiubli@xxxxxxxxxx> Reviewed-by: Milind Changire <mchangir@xxxxxxxxxx> Signed-off-by: Ilya Dryomov <idryomov@xxxxxxxxx> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- fs/ceph/mds_client.c | 4 ++-- fs/ceph/mds_client.h | 5 +++++ fs/ceph/super.c | 10 ++++++++++ 3 files changed, 17 insertions(+), 2 deletions(-) --- a/fs/ceph/mds_client.c +++ b/fs/ceph/mds_client.c @@ -4607,7 +4607,7 @@ static void delayed_work(struct work_str dout("mdsc delayed_work\n"); - if (mdsc->stopping) + if (mdsc->stopping >= CEPH_MDSC_STOPPING_FLUSHED) return; mutex_lock(&mdsc->mutex); @@ -4786,7 +4786,7 @@ void send_flush_mdlog(struct ceph_mds_se void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc) { dout("pre_umount\n"); - mdsc->stopping = 1; + mdsc->stopping = CEPH_MDSC_STOPPING_BEGIN; ceph_mdsc_iterate_sessions(mdsc, send_flush_mdlog, true); ceph_mdsc_iterate_sessions(mdsc, lock_unlock_session, false); --- a/fs/ceph/mds_client.h +++ b/fs/ceph/mds_client.h @@ -370,6 +370,11 @@ struct cap_wait { int want; }; +enum { + CEPH_MDSC_STOPPING_BEGIN = 1, + CEPH_MDSC_STOPPING_FLUSHED = 2, +}; + /* * mds client state */ --- a/fs/ceph/super.c +++ b/fs/ceph/super.c @@ -1227,6 +1227,16 @@ static void ceph_kill_sb(struct super_bl ceph_mdsc_pre_umount(fsc->mdsc); flush_fs_workqueues(fsc); + /* + * Though the kill_anon_super() will finally trigger the + * sync_filesystem() anyway, we still need to do it here + * and then bump the stage of shutdown to stop the work + * queue as earlier as possible. + */ + sync_filesystem(s); + + fsc->mdsc->stopping = CEPH_MDSC_STOPPING_FLUSHED; + kill_anon_super(s); fsc->client->extra_mon_dispatch = NULL;
