cluster/afr: Heal directory rename without rmdir/mkdir

Problem1:
When a directory is renamed while a brick
is down entry-heal always did an rm -rf on that directory on
the sink on old location and did mkdir and created the directory
hierarchy again in the new location. This is inefficient.

Problem2:
Renamedir heal order may lead to a scenario where directory in
the new location could be created before deleting it from old
location leading to 2 directories with same gfid in posix.

Fix:
As part of heal, if oldlocation is healed first and is not present in
source-brick always rename it into a hidden directory inside the
sink-brick so that when heal is triggered in new-location shd can
rename it from this hidden directory to the new-location.

If new-location heal is triggered first and it detects that the
directory already exists in the brick, then it should skip healing the
directory until it appears in the hidden directory.

Credits: Ravi for rename-data-loss.t script

Fixes: #1211
Change-Id: I0cba2006f35cd03d314d18211ce0bd530e254843
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>

1 files changed, 176 insertions, 2 deletions

diff --git a/xlators/cluster/afr/src/afr-self-heald.c b/xlators/cluster/afr/src/afr-self-heald.c
index f2e08908c24..109fd4b7421 100644
--- a/xlators/cluster/afr/src/afr-self-heald.c
+++ b/xlators/cluster/afr/src/afr-self-heald.c
@@ -222,7 +222,7 @@ out:
 }
 
 int
-afr_shd_index_purge(xlator_t *subvol, inode_t *inode, char *name,
+afr_shd_entry_purge(xlator_t *subvol, inode_t *inode, char *name,
                     ia_type_t type)
 {
     int ret = 0;
@@ -424,7 +424,7 @@ afr_shd_index_heal(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
     ret = afr_shd_selfheal(healer, healer->subvol, gfid);
 
     if (ret == -ENOENT || ret == -ESTALE)
-        afr_shd_index_purge(subvol, parent->inode, entry->d_name, val);
+        afr_shd_entry_purge(subvol, parent->inode, entry->d_name, val);
 
     if (ret == 2)
         /* If bricks crashed in pre-op after creating indices/xattrop
@@ -843,6 +843,176 @@ out:
     return need_heal;
 }
 
+static int
+afr_shd_anon_inode_cleaner(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
+                           void *data)
+{
+    struct subvol_healer *healer = data;
+    afr_private_t *priv = healer->this->private;
+    call_frame_t *frame = NULL;
+    afr_local_t *local = NULL;
+    int ret = 0;
+    loc_t loc = {0};
+    int count = 0;
+    int i = 0;
+    int op_errno = 0;
+    struct iatt *iatt = NULL;
+    gf_boolean_t multiple_links = _gf_false;
+    unsigned char *gfid_present = alloca0(priv->child_count);
+    unsigned char *entry_present = alloca0(priv->child_count);
+    char *type = "file";
+
+    frame = afr_frame_create(healer->this, &ret);
+    if (!frame) {
+        ret = -ret;
+        goto out;
+    }
+    local = frame->local;
+    if (AFR_COUNT(local->child_up, priv->child_count) != priv->child_count) {
+        gf_msg_debug(healer->this->name, 0,
+                     "Not all bricks are up. Skipping "
+                     "cleanup of %s on %s",
+                     entry->d_name, subvol->name);
+        ret = 0;
+        goto out;
+    }
+
+    loc.inode = inode_new(parent->inode->table);
+    if (!loc.inode) {
+        ret = -ENOMEM;
+        goto out;
+    }
+    ret = gf_uuid_parse(entry->d_name, loc.gfid);
+    if (ret) {
+        ret = 0;
+        goto out;
+    }
+    AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup, &loc,
+               NULL);
+    for (i = 0; i < priv->child_count; i++) {
+        if (local->replies[i].op_ret == 0) {
+            count++;
+            gfid_present[i] = 1;
+            iatt = &local->replies[i].poststat;
+            if (iatt->ia_type == IA_IFDIR) {
+                type = "dir";
+            }
+
+            if (i == healer->subvol) {
+                if (local->replies[i].poststat.ia_nlink > 1) {
+                    multiple_links = _gf_true;
+                }
+            }
+        } else if (local->replies[i].op_errno != ENOENT &&
+                   local->replies[i].op_errno != ESTALE) {
+            /*We don't have complete view. Skip the entry*/
+            gf_msg_debug(healer->this->name, local->replies[i].op_errno,
+                         "Skipping cleanup of %s on %s", entry->d_name,
+                         subvol->name);
+            ret = 0;
+            goto out;
+        }
+    }
+
+    /*Inode is deleted from subvol*/
+    if (count == 1 || (iatt->ia_type != IA_IFDIR && multiple_links)) {
+        gf_msg(healer->this->name, GF_LOG_WARNING, 0,
+               AFR_MSG_EXPUNGING_FILE_OR_DIR, "expunging %s %s/%s on %s", type,
+               priv->anon_inode_name, entry->d_name, subvol->name);
+        ret = afr_shd_entry_purge(subvol, parent->inode, entry->d_name,
+                                  iatt->ia_type);
+        if (ret == -ENOENT || ret == -ESTALE)
+            ret = 0;
+    } else if (count > 1) {
+        loc_wipe(&loc);
+        loc.parent = inode_ref(parent->inode);
+        loc.name = entry->d_name;
+        loc.inode = inode_new(parent->inode->table);
+        if (!loc.inode) {
+            ret = -ENOMEM;
+            goto out;
+        }
+        AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup,
+                   &loc, NULL);
+        count = 0;
+        for (i = 0; i < priv->child_count; i++) {
+            if (local->replies[i].op_ret == 0) {
+                count++;
+                entry_present[i] = 1;
+                iatt = &local->replies[i].poststat;
+            } else if (local->replies[i].op_errno != ENOENT &&
+                       local->replies[i].op_errno != ESTALE) {
+                /*We don't have complete view. Skip the entry*/
+                gf_msg_debug(healer->this->name, local->replies[i].op_errno,
+                             "Skipping cleanup of %s on %s", entry->d_name,
+                             subvol->name);
+                ret = 0;
+                goto out;
+            }
+        }
+        for (i = 0; i < priv->child_count; i++) {
+            if (gfid_present[i] && !entry_present[i]) {
+                /*Entry is not anonymous on at least one subvol*/
+                gf_msg_debug(healer->this->name, 0,
+                             "Valid entry present on %s "
+                             "Skipping cleanup of %s on %s",
+                             priv->children[i]->name, entry->d_name,
+                             subvol->name);
+                ret = 0;
+                goto out;
+            }
+        }
+
+        gf_msg(healer->this->name, GF_LOG_WARNING, 0,
+               AFR_MSG_EXPUNGING_FILE_OR_DIR,
+               "expunging %s %s/%s on all subvols", type, priv->anon_inode_name,
+               entry->d_name);
+        ret = 0;
+        for (i = 0; i < priv->child_count; i++) {
+            op_errno = -afr_shd_entry_purge(priv->children[i], loc.parent,
+                                            entry->d_name, iatt->ia_type);
+            if (op_errno != ENOENT && op_errno != ESTALE) {
+                ret |= -op_errno;
+            }
+        }
+    }
+
+out:
+    if (frame)
+        AFR_STACK_DESTROY(frame);
+    loc_wipe(&loc);
+    return ret;
+}
+
+static void
+afr_cleanup_anon_inode_dir(struct subvol_healer *healer)
+{
+    int ret = 0;
+    call_frame_t *frame = NULL;
+    afr_private_t *priv = healer->this->private;
+    loc_t loc = {0};
+
+    ret = afr_anon_inode_create(healer->this, healer->subvol, &loc.inode);
+    if (ret)
+        goto out;
+
+    frame = afr_frame_create(healer->this, &ret);
+    if (!frame) {
+        ret = -ret;
+        goto out;
+    }
+
+    ret = syncop_mt_dir_scan(frame, priv->children[healer->subvol], &loc,
+                             GF_CLIENT_PID_SELF_HEALD, healer,
+                             afr_shd_anon_inode_cleaner, NULL,
+                             priv->shd.max_threads, priv->shd.wait_qlength);
+out:
+    if (frame)
+        AFR_STACK_DESTROY(frame);
+    loc_wipe(&loc);
+    return;
+}
+
 void *
 afr_shd_index_healer(void *data)
 {
@@ -900,6 +1070,10 @@ afr_shd_index_healer(void *data)
             sleep(1);
         } while (ret > 0);
 
+        if (ret == 0) {
+            afr_cleanup_anon_inode_dir(healer);
+        }
+
         if (ret == 0 && pre_crawl_xdata &&
             !healer->crawl_event.heal_failed_count) {
             afr_shd_ta_check_and_unset_xattrs(this, &loc, healer,