On Mon, 2019-06-17 at 20:55 +0800, Yan, Zheng wrote:
> When remounting aborted mount, also reset client's entity addr.
> 'umount -f /ceph; mount -o remount /ceph' can be used for recovering
> from blacklist.
>
Why do I need to umount here? Once the filesystem is unmounted, then the
'-o remount' becomes superfluous, no? In fact, I get an error back when
I try to remount an unmounted filesystem:
$ sudo umount -f /mnt/cephfs ; sudo mount -o remount /mnt/cephfs
mount: /mnt/cephfs: mount point not mounted or bad option.
My client isn't blacklisted above, so I guess you're counting on the
umount returning without having actually unmounted the filesystem?
I think this ought to not need a umount first. From a UI standpoint,
just doing a "mount -o remount" ought to be sufficient to clear this.
Also, how would an admin know that this is something they ought to try?
Is there a way for them to know that their client has been blacklisted?
> Signed-off-by: "Yan, Zheng" <zyan@xxxxxxxxxx>
> ---
> fs/ceph/mds_client.c | 16 +++++++++++++---
> fs/ceph/super.c | 23 +++++++++++++++++++++--
> 2 files changed, 34 insertions(+), 5 deletions(-)
>
> diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
> index 19c62cf7d5b8..188c33709d9a 100644
> --- a/fs/ceph/mds_client.c
> +++ b/fs/ceph/mds_client.c
> @@ -1378,9 +1378,12 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
> struct ceph_cap_flush *cf;
> struct ceph_mds_client *mdsc = fsc->mdsc;
>
> - if (ci->i_wrbuffer_ref > 0 &&
> - READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
> - invalidate = true;
> + if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
> + if (inode->i_data.nrpages > 0)
> + invalidate = true;
> + if (ci->i_wrbuffer_ref > 0)
> + mapping_set_error(&inode->i_data, -EIO);
> + }
>
> while (!list_empty(&ci->i_cap_flush_list)) {
> cf = list_first_entry(&ci->i_cap_flush_list,
> @@ -4350,7 +4353,12 @@ void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
> session = __ceph_lookup_mds_session(mdsc, mds);
> if (!session)
> continue;
> +
> + if (session->s_state == CEPH_MDS_SESSION_REJECTED)
> + __unregister_session(mdsc, session);
> + __wake_requests(mdsc, &session->s_waiting);
> mutex_unlock(&mdsc->mutex);
> +
> mutex_lock(&session->s_mutex);
> __close_session(mdsc, session);
> if (session->s_state == CEPH_MDS_SESSION_CLOSING) {
> @@ -4359,9 +4367,11 @@ void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
> }
> mutex_unlock(&session->s_mutex);
> ceph_put_mds_session(session);
> +
> mutex_lock(&mdsc->mutex);
> kick_requests(mdsc, mds);
> }
> +
> __wake_requests(mdsc, &mdsc->waiting_for_map);
> mutex_unlock(&mdsc->mutex);
> }
> diff --git a/fs/ceph/super.c b/fs/ceph/super.c
> index 67eb9d592ab7..a6a3c065f697 100644
> --- a/fs/ceph/super.c
> +++ b/fs/ceph/super.c
> @@ -833,8 +833,27 @@ static void ceph_umount_begin(struct super_block *sb)
>
> static int ceph_remount(struct super_block *sb, int *flags, char *data)
> {
> - sync_filesystem(sb);
> - return 0;
> + struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
> +
> + if (fsc->mount_state != CEPH_MOUNT_SHUTDOWN) {
> + sync_filesystem(sb);
> + return 0;
> + }
> +
> + /* Make sure all page caches get invalidated.
> + * see remove_session_caps_cb() */
> + flush_workqueue(fsc->inode_wq);
> + /* In case that we were blacklisted. This also reset
> + * all mon/osd connections */
> + ceph_reset_client_addr(fsc->client);
> +
> + ceph_osdc_clear_abort_err(&fsc->client->osdc);
> + fsc->mount_state = 0;
> +
> + if (!sb->s_root)
> + return 0;
> + return __ceph_do_getattr(d_inode(sb->s_root), NULL,
> + CEPH_STAT_CAP_INODE, true);
> }
>
> static const struct super_operations ceph_super_ops = {
--
Jeff Layton <jlayton@xxxxxxxxxx>
