ceph: resubmit requests on pg mapping change (not just primary change)

OSD requests need to be resubmitted on any pg mapping change, not just when
the pg primary changes.  Resending only when the primary changes results in
occasional 'hung' requests during osd cluster recovery or rebalancing.

Signed-off-by: Sage Weil <sage@newdream.net>

1 files changed, 15 insertions, 4 deletions

diff --git a/fs/ceph/osd_client.c b/fs/ceph/osd_client.c
index 8128082a028e..3514f71ff85f 100644
--- a/fs/ceph/osd_client.c
+++ b/fs/ceph/osd_client.c
@@ -565,7 +565,8 @@ static int __map_osds(struct ceph_osd_client *osdc,
 {
         struct ceph_osd_request_head *reqhead = req->r_request->front.iov_base;
         struct ceph_pg pgid;
-        int o = -1;
+        int acting[CEPH_PG_MAX_SIZE];
+        int o = -1, num = 0;
         int err;
 
         dout("map_osds %p tid %lld\n", req, req->r_tid);
@@ -576,10 +577,16 @@ static int __map_osds(struct ceph_osd_client *osdc,
         pgid = reqhead->layout.ol_pgid;
         req->r_pgid = pgid;
 
-        o = ceph_calc_pg_primary(osdc->osdmap, pgid);
+        err = ceph_calc_pg_acting(osdc->osdmap, pgid, acting);
+        if (err > 0) {
+                o = acting[0];
+                num = err;
+        }
 
         if ((req->r_osd && req->r_osd->o_osd == o &&
-             req->r_sent >= req->r_osd->o_incarnation) ||
+             req->r_sent >= req->r_osd->o_incarnation &&
+             req->r_num_pg_osds == num &&
+             memcmp(req->r_pg_osds, acting, sizeof(acting[0])*num) == 0) ||
             (req->r_osd == NULL && o == -1))
                 return 0;  /* no change */
 
@@ -587,6 +594,10 @@ static int __map_osds(struct ceph_osd_client *osdc,
              req->r_tid, le32_to_cpu(pgid.pool), le16_to_cpu(pgid.ps), o,
              req->r_osd ? req->r_osd->o_osd : -1);
 
+        /* record full pg acting set */
+        memcpy(req->r_pg_osds, acting, sizeof(acting[0]) * num);
+        req->r_num_pg_osds = num;
+
         if (req->r_osd) {
                 __cancel_request(req);
                 list_del_init(&req->r_osd_item);
@@ -612,7 +623,7 @@ static int __map_osds(struct ceph_osd_client *osdc,
                 __remove_osd_from_lru(req->r_osd);
                 list_add(&req->r_osd_item, &req->r_osd->o_requests);
         }
-        err = 1;   /* osd changed */
+        err = 1;   /* osd or pg changed */
 
 out:
         return err;