cluster/afr : Prevent excessive logging of split-brain messages.

        Running the volume heal info command would result in
excessive logging of split-brain messages. After this patch,
running heal info command will not log the split brain messages.
This info is now displayed in the output of heal info command
instead.
If a file is in split-brain, a message "Is in split-brain"
will be written against its name.

Change-Id: Ib8979be04f5ac7c59ce3ad1185886bb54b8be808
BUG: 1161102
Signed-off-by: Anuradha <atalur@redhat.com>
Reviewed-on: http://review.gluster.org/9069
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Niels de Vos <ndevos@redhat.com>

6 files changed, 37 insertions, 10 deletions

diff --git a/heal/src/glfs-heal.c b/heal/src/glfs-heal.c
index 78716ad64da..236361c2605 100644
--- a/heal/src/glfs-heal.c
+++ b/heal/src/glfs-heal.c
@@ -141,6 +141,20 @@ _is_possibly_healing (dict_t *xattr_rsp)
         return _gf_false;
 }
 
+static gf_boolean_t
+_is_in_split_brain (dict_t *xattr_rsp)
+{
+        int     ret = 0;
+        int     spb = 0;
+
+        ret = dict_get_int32 (xattr_rsp, "split-brain", &spb);
+        if ((ret == 0) && spb) {
+                return _gf_true;
+        }
+
+        return _gf_false;
+}
+
 #define RESET_ENTRIES(loc, shf, ope, pth, rsp, grsp) \
         do {                                    \
                 loc_wipe (&loc);                \
@@ -240,9 +254,10 @@ glfsh_process_entries (xlator_t *xl, loc_t *parentloc, gf_dirent_t *entries,
                         continue;
                 }
 
-                ret = dict_get_int32 (xattr_rsp, "sh-failed", &sh_failed);
-
-                sh_pending = _is_self_heal_pending (xattr_rsp);
+                if (xattr_rsp) {
+                        ret = dict_get_int32 (xattr_rsp, "sh-failed", &sh_failed);
+                        sh_pending = _is_self_heal_pending (xattr_rsp);
+                }
                 //File/dir is undergoing I/O
                 if (!op_errno && !sh_failed && !sh_pending)
                         continue;
@@ -260,7 +275,10 @@ glfsh_process_entries (xlator_t *xl, loc_t *parentloc, gf_dirent_t *entries,
                 ret = dict_get_str (getxattr_rsp, GFID_TO_PATH_KEY, &path);
 
                 (*num_entries)++;
-                if (_is_possibly_healing (xattr_rsp)) {
+                if (xattr_rsp && _is_in_split_brain (xattr_rsp)) {
+                        printf ("%s - Is in split-brain\n",
+                                path ? path : uuid_utoa (entry_loc.gfid));
+                } else if (xattr_rsp && _is_possibly_healing (xattr_rsp)) {
                         printf ("%s - Possibly undergoing heal\n",
                                 path ? path : uuid_utoa (entry_loc.gfid));
                 } else {
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index fae3d4c1731..59b8038dcac 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -1923,6 +1923,12 @@ afr_self_heal_lookup_unwind (call_frame_t *frame, xlator_t *this,
         if (!xattr)
                 goto out;
 
+        if (afr_is_split_brain (this, local->cont.lookup.inode)) {
+                ret = dict_set_int32 (xattr, "split-brain", 1);
+                if (ret)
+                        gf_log (this->name, GF_LOG_ERROR, "%s: Failed to set "
+                                "split-brain to true", local->loc.path);
+        }
         if (sh_failed) {
                 ret = dict_set_int32 (xattr, "sh-failed", sh_failed);
                 if (ret)
diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index 5f47a4cc410..d50914fa103 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -216,12 +216,15 @@ out:
 
 void
 afr_sh_print_split_brain_log (int32_t *pending_matrix[], xlator_t *this,
-                              const char *loc, afr_spb_state_t mdata,
+                              afr_local_t *local, afr_spb_state_t mdata,
                               afr_spb_state_t data)
 {
         char *buf      = NULL;
         char *free_ptr = NULL;
 
+        if (local->self_heal.dry_run)
+                return;
+
         buf = afr_get_pending_matrix_str (pending_matrix, this);
         if (buf)
                 free_ptr = buf;
@@ -231,8 +234,8 @@ afr_sh_print_split_brain_log (int32_t *pending_matrix[], xlator_t *this,
 
         gf_log (this->name, GF_LOG_ERROR, "Unable to self-heal contents of '%s'"
                 " (possible %s split-brain). Please delete the file from all but "
-                "the preferred subvolume.%s", loc, (mdata == SPB) ? "metadata" :
-                (data == SPB) ? "data" : "", buf);
+                "the preferred subvolume.%s", local->loc.path, (mdata == SPB) ?
+                "metadata" : (data == SPB) ? "data" : "", buf);
         GF_FREE (free_ptr);
         return;
 }
diff --git a/xlators/cluster/afr/src/afr-self-heal-common.h b/xlators/cluster/afr/src/afr-self-heal-common.h
index 4c926112dc7..8c01168d9ce 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.h
+++ b/xlators/cluster/afr/src/afr-self-heal-common.h
@@ -30,7 +30,7 @@ afr_sh_print_pending_matrix (int32_t *pending_matrix[], xlator_t *this);
 
 void
 afr_sh_print_split_brain_log (int32_t *pending_matrix[], xlator_t *this,
-                              const char *loc, afr_spb_state_t mdata,
+                              afr_local_t *local, afr_spb_state_t mdata,
                               afr_spb_state_t data);
 
 int
diff --git a/xlators/cluster/afr/src/afr-self-heal-data.c b/xlators/cluster/afr/src/afr-self-heal-data.c
index 712ca5561f1..f85e7247be4 100644
--- a/xlators/cluster/afr/src/afr-self-heal-data.c
+++ b/xlators/cluster/afr/src/afr-self-heal-data.c
@@ -1014,7 +1014,7 @@ afr_sh_data_fxattrop_fstat_done (call_frame_t *frame, xlator_t *this)
 
         if (nsources == -1) {
                 afr_sh_print_split_brain_log (sh->pending_matrix, this,
-                                              local->loc.path, DONT_KNOW,
+                                              local, DONT_KNOW,
                                               SPB);
                 afr_set_split_brain (this, sh->inode, DONT_KNOW, SPB);
 
diff --git a/xlators/cluster/afr/src/afr-self-heal-metadata.c b/xlators/cluster/afr/src/afr-self-heal-metadata.c
index 2e8c076fe42..829a8c35da6 100644
--- a/xlators/cluster/afr/src/afr-self-heal-metadata.c
+++ b/xlators/cluster/afr/src/afr-self-heal-metadata.c
@@ -486,7 +486,7 @@ afr_sh_metadata_fix (call_frame_t *frame, xlator_t *this,
 
         if (nsources == -1) {
                 afr_sh_print_split_brain_log (sh->pending_matrix, this,
-                                              local->loc.path, SPB,
+                                              local, SPB,
                                               DONT_KNOW);
                 afr_set_split_brain (this, sh->inode, SPB, DONT_KNOW);
                 afr_sh_metadata_fail (frame, this);