From e1a4541ec0b951685a49d1f72d183681e6433a45 Mon Sep 17 00:00:00 2001 From: Xiubo Li Date: Mon, 5 Jul 2021 09:22:57 +0800 Subject: [PATCH] ceph: flush the mdlog before waiting on unsafe reqs For the client requests who will have unsafe and safe replies from MDS daemons, in the MDS side the MDS daemons won't flush the mdlog (journal log) immediatelly, because they think it's unnecessary. That's true for most cases but not all, likes the fsync request. The fsync will wait until all the unsafe replied requests to be safely replied. Normally if there have multiple threads or clients are running, the whole mdlog in MDS daemons could be flushed in time if any request will trigger the mdlog submit thread. So usually we won't experience the normal operations will stuck for a long time. But in case there has only one client with only thread is running, the stuck phenomenon maybe obvious and the worst case it must wait at most 5 seconds to wait the mdlog to be flushed by the MDS's tick thread periodically. This patch will trigger to flush the mdlog in the relevant and auth MDSes to which the in-flight requests are sent just before waiting the unsafe requests to finish. Signed-off-by: Xiubo Li Reviewed-by: Jeff Layton Signed-off-by: Ilya Dryomov --- fs/ceph/caps.c | 76 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 76 insertions(+) diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index a42dbc343749..d5a46556e20e 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -2219,6 +2219,7 @@ static int caps_are_flushed(struct inode *inode, u64 flush_tid) */ static int unsafe_request_wait(struct inode *inode) { + struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc; struct ceph_inode_info *ci = ceph_inode(inode); struct ceph_mds_request *req1 = NULL, *req2 = NULL; int ret, err = 0; @@ -2238,6 +2239,81 @@ static int unsafe_request_wait(struct inode *inode) } spin_unlock(&ci->i_unsafe_lock); + /* + * Trigger to flush the journal logs in all the relevant MDSes + * manually, or in the worst case we must wait at most 5 seconds + * to wait the journal logs to be flushed by the MDSes periodically. + */ + if (req1 || req2) { + struct ceph_mds_session **sessions = NULL; + struct ceph_mds_session *s; + struct ceph_mds_request *req; + unsigned int max; + int i; + + /* + * The mdsc->max_sessions is unlikely to be changed + * mostly, here we will retry it by reallocating the + * sessions arrary memory to get rid of the mdsc->mutex + * lock. + */ +retry: + max = mdsc->max_sessions; + sessions = krealloc(sessions, max * sizeof(s), __GFP_ZERO); + if (!sessions) + return -ENOMEM; + + spin_lock(&ci->i_unsafe_lock); + if (req1) { + list_for_each_entry(req, &ci->i_unsafe_dirops, + r_unsafe_dir_item) { + s = req->r_session; + if (unlikely(s->s_mds > max)) { + spin_unlock(&ci->i_unsafe_lock); + goto retry; + } + if (!sessions[s->s_mds]) { + s = ceph_get_mds_session(s); + sessions[s->s_mds] = s; + } + } + } + if (req2) { + list_for_each_entry(req, &ci->i_unsafe_iops, + r_unsafe_target_item) { + s = req->r_session; + if (unlikely(s->s_mds > max)) { + spin_unlock(&ci->i_unsafe_lock); + goto retry; + } + if (!sessions[s->s_mds]) { + s = ceph_get_mds_session(s); + sessions[s->s_mds] = s; + } + } + } + spin_unlock(&ci->i_unsafe_lock); + + /* the auth MDS */ + spin_lock(&ci->i_ceph_lock); + if (ci->i_auth_cap) { + s = ci->i_auth_cap->session; + if (!sessions[s->s_mds]) + sessions[s->s_mds] = ceph_get_mds_session(s); + } + spin_unlock(&ci->i_ceph_lock); + + /* send flush mdlog request to MDSes */ + for (i = 0; i < max; i++) { + s = sessions[i]; + if (s) { + send_flush_mdlog(s); + ceph_put_mds_session(s); + } + } + kfree(sessions); + } + dout("unsafe_request_wait %p wait on tid %llu %llu\n", inode, req1 ? req1->r_tid : 0ULL, req2 ? req2->r_tid : 0ULL); if (req1) {