afr: heal gfid as a part of entry heal

Problem:
If a brick crashes after an entry (file or dir) is created but before
gfid is assigned, the good bricks will have pending entry heal xattrs
but the heal won't complete because afr_selfheal_recreate_entry() tries
to create the entry again and it fails with EEXIST.

Fix:
We could have fixed posx_mknod/mkdir etc to assign the gfid if the file
already exists but the right thing to do seems to be to trigger a lookup
on the bad brick and let it heal the gfid instead of winding an
mknod/mkdir in the first place.

Change-Id: I82f76665a7541f1893ef8d847b78af6466aff1ff
BUG: 1493415
Signed-off-by: Ravishankar N <ravishankar@redhat.com>

1 files changed, 78 insertions, 0 deletions

diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index 55b978bf962..20e81dd43e0 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -20,6 +20,84 @@ void
 afr_heal_synctask (xlator_t *this, afr_local_t *local);
 
 int
+afr_lookup_and_heal_gfid (xlator_t *this, inode_t *parent, const char *name,
+                          inode_t *inode, struct afr_reply *replies,
+                          int source,  void *gfid)
+{
+        afr_private_t *priv = NULL;
+        call_frame_t   *frame    = NULL;
+        afr_local_t *local = NULL;
+        unsigned char *wind_on = NULL;
+        ia_type_t ia_type = IA_INVAL;
+        dict_t *xdata = NULL;
+        loc_t loc = {0, };
+        int ret = 0;
+        int i = 0;
+
+        priv = this->private;
+        wind_on = alloca0 (priv->child_count);
+        ia_type = replies[source].poststat.ia_type;
+
+        /* gfid heal on those subvolumes that do not have gfid associated
+         * with the inode and update those replies.
+         */
+        for (i = 0; i < priv->child_count; i++) {
+                if (!replies[i].valid || replies[i].op_ret != 0)
+                        continue;
+                if (!gf_uuid_is_null (replies[i].poststat.ia_gfid) ||
+                    replies[i].poststat.ia_type != ia_type)
+                        continue;
+
+                wind_on[i] = 1;
+        }
+
+        if (AFR_COUNT(wind_on, priv->child_count) == 0)
+                return 0;
+
+        xdata = dict_new ();
+        if (!xdata) {
+                ret = -ENOMEM;
+                goto out;
+        }
+
+        ret = dict_set_static_bin (xdata, "gfid-req", gfid, 16);
+        if (ret) {
+                ret = -ENOMEM;
+                goto out;
+        }
+
+        frame = afr_frame_create (this);
+        if (!frame) {
+                ret = -ENOMEM;
+                goto out;
+        }
+
+        local = frame->local;
+        loc.parent = inode_ref (parent);
+        gf_uuid_copy (loc.pargfid, parent->gfid);
+        loc.name = name;
+        loc.inode = inode_ref (inode);
+
+        AFR_ONLIST (wind_on, frame, afr_selfheal_discover_cbk, lookup,
+                    &loc, xdata);
+
+        for (i = 0; i < priv->child_count; i++) {
+                if (!wind_on[i])
+                        continue;
+                afr_reply_wipe (&replies[i]);
+                afr_reply_copy (&replies[i], &local->replies[i]);
+        }
+out:
+        loc_wipe (&loc);
+        if (frame)
+                AFR_STACK_DESTROY (frame);
+        if (xdata)
+                dict_unref (xdata);
+
+        return ret;
+}
+
+int
 afr_gfid_sbrain_source_from_src_brick (xlator_t *this,
                                        struct afr_reply *replies,
                                        char *src_brick)