dht: "replica.split-brain-status" attribute value is not correct

Problem: In a distributed-replicate volume attribute
         "replica.split-brain-status" value does not display split-brain
           condition though directory is in split-brain.
         If directory is in split brain on mutiple replica-pairs
         it does not show full list of replica pairs.

Solution: Update the dht_aggregate code to aggregate the xattr
          value in this specific condition.

Fix:      1) function getChoices returns the choices from split-brain
             status string.
          2) function add_opt adding the choices to local buffer to
             store in dictionary
          3) For the key "replica.split-brain-status" function dht_aggregate
             call dht_aggregate_split_brain_xattr to prepare the list.

Test:     To verify the patch followed below steps
          1) Create a distributed replica volume and create mount point
          2) Stop heal daemon
          3) Touch file and directories on mount point
             mkdir test{1..5};touch tmp{1..5}
          4) Down brick process on one of the replica set
             pkill -9 glusterfsd
          5) Change permission of dir on mount point
             chmod 755 test{1..5}
          6) Restart brick process on node with force option
          7) kill brick process on other node in same replica set
          8) Change permission of dir again on mount point
             chmod 766 test{1..5}
          9) Reexecute same step from 4-9 on other replica set also
          10) After check heal status on server it will show dir's are
              in split brain on all replica sets
          11) After check the replica.split-brain-status attr on mount
              point it will show wrong status of split brain.
          12) After apply the patch the attribute shows correct value.

> Change-Id: Icdfd72005a4aa82337c342762775a3d1761bbe4a
> Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
> Reviewed-on: http://review.gluster.org/15201
> Smoke: Gluster Build System <jenkins@build.gluster.org>
> NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
> CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
> Reviewed-by: Raghavendra G <rgowdapp@redhat.com>
> (cherry picked from commit c4e9ec653c946002ab6d4c71ee8e6df056438a04)

Change-Id: I85a5ae60189066d9e80799f00f1352c2f33ef4f8
Backport of commit c4e9ec653c946002ab6d4c71ee8e6df056438a04
BUG: 1375098
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
Reviewed-on: http://review.gluster.org/15467
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Raghavendra G <rgowdapp@redhat.com>

3 files changed, 294 insertions, 12 deletions

diff --git a/tests/bugs/bug-1368312.t b/tests/bugs/bug-1368312.t
new file mode 100644
index 00000000000..135048f448e
--- /dev/null
+++ b/tests/bugs/bug-1368312.t
@@ -0,0 +1,84 @@
+#!/bin/bash
+. $(dirname $0)/../include.rc
+. $(dirname $0)/../volume.rc
+cleanup;
+
+function compare_get_split_brain_status {
+        local path=$1
+        local choice=$2
+        echo `getfattr -n replica.split-brain-status $path` | cut -f2 -d"=" | sed -e 's/^"//'  -e 's/"$//' | grep $choice
+        if [ $? -ne 0 ]
+        then
+                echo 1
+        else
+                echo 0
+        fi
+
+}
+
+TEST glusterd
+TEST pidof glusterd
+TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{0,1,2,3,4,5}
+TEST $CLI volume start $V0
+
+#Disable self-heal-daemon
+TEST $CLI volume set $V0 cluster.self-heal-daemon off
+
+TEST glusterfs --volfile-id=$V0 --volfile-server=$H0 --entry-timeout=0 $M0;
+
+TEST mkdir $M0/tmp1
+
+#Create metadata split-brain
+TEST kill_brick $V0 $H0 $B0/${V0}0
+TEST chmod 666 $M0/tmp1
+TEST $CLI volume start $V0 force
+TEST kill_brick $V0 $H0 $B0/${V0}1
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+
+TEST chmod 757 $M0/tmp1
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 0
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 1
+
+EXPECT 2 get_pending_heal_count $V0
+
+
+TEST kill_brick $V0 $H0 $B0/${V0}2
+TEST chmod 755 $M0/tmp1
+TEST $CLI volume start $V0 force
+TEST kill_brick $V0 $H0 $B0/${V0}3
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 2
+
+TEST chmod 766 $M0/tmp1
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 2
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 3
+
+EXPECT 4 get_pending_heal_count $V0
+
+TEST kill_brick $V0 $H0 $B0/${V0}4
+TEST chmod 765 $M0/tmp1
+TEST $CLI volume start $V0 force
+TEST kill_brick $V0 $H0 $B0/${V0}5
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 4
+
+TEST chmod 756 $M0/tmp1
+
+TEST $CLI volume start $V0 force
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 4
+EXPECT_WITHIN $PROCESS_UP_TIMEOUT "1" afr_child_up_status $V0 5
+
+EXPECT 6 get_pending_heal_count $V0
+
+cd $M0
+EXPECT 0 compare_get_split_brain_status ./tmp1 patchy-client-0
+EXPECT 0 compare_get_split_brain_status ./tmp1 patchy-client-1
+EXPECT 0 compare_get_split_brain_status ./tmp1 patchy-client-2
+EXPECT 0 compare_get_split_brain_status ./tmp1 patchy-client-3
+EXPECT 0 compare_get_split_brain_status ./tmp1 patchy-client-4
+EXPECT 0 compare_get_split_brain_status ./tmp1 patchy-client-5
+
+cd -
+cleanup
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index 2eb44a1dd82..776dba0ee94 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -126,6 +126,190 @@ out:
 }
 
 
+int add_opt(char **optsp, const char *opt)
+{
+        char *newopts = NULL;
+        unsigned oldsize = 0;
+        unsigned newsize = 0;
+
+        if (*optsp == NULL)
+                newopts = gf_strdup (opt);
+        else {
+                oldsize = strlen (*optsp);
+                newsize = oldsize + 1 + strlen (opt) + 1;
+                newopts = GF_REALLOC (*optsp, newsize);
+                if (newopts)
+                        sprintf (newopts + oldsize, ",%s", opt);
+        }
+        if (newopts == NULL) {
+                gf_msg ("dht", GF_LOG_WARNING, 0,
+                        DHT_MSG_NO_MEMORY,
+                        "Error to add choices in buffer in add_opt");
+                return -1;
+        }
+        *optsp = newopts;
+        return 0;
+}
+
+/* Return Choice list from Split brain status */
+char *
+getChoices (const char *value)
+{
+        int i = 0;
+        char *ptr = NULL;
+        char *tok = NULL;
+        char *result = NULL;
+        char *newval = NULL;
+
+        ptr = strstr (value, "Choices:");
+        if (!ptr) {
+                result = ptr;
+                goto out;
+        }
+
+        newval = gf_strdup (ptr);
+        if (!newval) {
+                result = newval;
+                goto out;
+        }
+
+        tok = strtok (newval, ":");
+        if (!tok) {
+                result = tok;
+                goto out;
+        }
+
+        while (tok) {
+                i++;
+                if (i == 2)
+                        break;
+                tok = strtok (NULL, ":");
+        }
+
+        result = gf_strdup (tok);
+
+out:
+        if (newval)
+                GF_FREE (newval);
+
+        return result;
+}
+
+/* This function prepare a list of choices for key
+   (replica.split-brain-status) in   case of metadata split brain
+   only on the basis of key-value passed to this function.
+   After prepare the list of choices it update the same key in dict
+   with this value to reflect the same in
+   replica.split-brain-status attr for file.
+
+*/
+
+int
+dht_aggregate_split_brain_xattr (dict_t *dst, char *key, data_t *value)
+{
+
+        int              ret            = 0;
+        char            *oldvalue       = NULL;
+        char            *old_choice     = NULL;
+        char            *new_choice     = NULL;
+        char            *full_choice    = NULL;
+        char            *status         = NULL;
+
+        if (value == NULL) {
+                gf_msg ("dht", GF_LOG_WARNING, 0,
+                        DHT_MSG_DATA_NULL,
+                        "GF_AFR_SBRAIN_STATUS value is NULL");
+                ret = -1;
+                goto out;
+        }
+
+        ret = dict_get_str (dst, key, &oldvalue);
+        if (ret)
+                goto out;
+
+        if (oldvalue && (strstr (oldvalue, "not"))) {
+                gf_msg_debug ("dht", 0,
+                              "Need to update split-brain status in dict");
+                ret = -1;
+                goto out;
+        }
+        if (oldvalue && (strstr (oldvalue, "metadata-split-brain:yes"))
+                     && (strstr (oldvalue, "data-split-brain:no"))) {
+                if (strstr (value->data, "not")) {
+                        gf_msg_debug ("dht", 0,
+                                      "No need to update split-brain status");
+                        ret = 0;
+                        goto out;
+                }
+                if (strstr (value->data, "yes") &&
+                        (strncmp (oldvalue, value->data, strlen(oldvalue)))) {
+                        old_choice = getChoices (oldvalue);
+                        if (!old_choice) {
+                                gf_msg ("dht", GF_LOG_WARNING, 0,
+                                        DHT_MSG_NO_MEMORY,
+                                        "Error to get choices");
+                                ret = -1;
+                                goto out;
+                        }
+
+                        ret = add_opt (&full_choice, old_choice);
+                        if (ret) {
+                                gf_msg ("dht", GF_LOG_WARNING, 0,
+                                         DHT_MSG_NO_MEMORY,
+                                         "Error to add choices");
+                                ret = -1;
+                                goto out;
+                        }
+
+                        new_choice = getChoices (value->data);
+                        if (!new_choice) {
+                                gf_msg ("dht", GF_LOG_WARNING, 0,
+                                        DHT_MSG_NO_MEMORY,
+                                        "Error to get choices");
+                                ret = -1;
+                                goto out;
+                        }
+
+                        ret = add_opt (&full_choice, new_choice);
+                        if (ret) {
+                                gf_msg ("dht", GF_LOG_WARNING, 0,
+                                       DHT_MSG_NO_MEMORY,
+                                       "Error to add choices ");
+                                ret = -1;
+                                goto out;
+                        }
+                        ret = gf_asprintf (&status,
+                                           "data-split-brain:%s    "
+                                           "metadata-split-brain:%s   Choices:%s",
+                                           "no", "yes", full_choice);
+
+                        if (-1 == ret) {
+                                gf_msg ("dht", GF_LOG_WARNING, 0,
+                                                 DHT_MSG_NO_MEMORY,
+                                                "Error to prepare status ");
+                                        goto out;
+                                }
+                        ret = dict_set_dynstr (dst, key, status);
+                        if (ret) {
+                                gf_msg ("dht", GF_LOG_WARNING, 0,
+                                        DHT_MSG_DICT_SET_FAILED,
+                                        "Failed to set full choice");
+                        }
+                }
+        }
+
+out:
+        if (old_choice)
+                GF_FREE (old_choice);
+        if (new_choice)
+                GF_FREE (new_choice);
+        if (full_choice)
+                GF_FREE (full_choice);
+
+        return ret;
+}
+
+
 
 int
 dht_aggregate (dict_t *this, char *key, data_t *value, void *data)
@@ -136,18 +320,22 @@ dht_aggregate (dict_t *this, char *key, data_t *value, void *data)
 
         dst = data;
 
-        if (strcmp (key, QUOTA_SIZE_KEY) == 0) {
+        /* compare split brain xattr only */
+        if (strcmp (key, GF_AFR_SBRAIN_STATUS) == 0) {
+                ret = dht_aggregate_split_brain_xattr(dst, key, value);
+                if (!ret)
+                        goto out;
+        } else if (strcmp (key, QUOTA_SIZE_KEY) == 0) {
                 ret = dht_aggregate_quota_xattr (dst, key, value);
                 if (ret) {
                         gf_msg ("dht", GF_LOG_WARNING, 0,
                                 DHT_MSG_AGGREGATE_QUOTA_XATTR_FAILED,
                                 "Failed to aggregate quota xattr");
-                        goto out;
                 }
+                goto out;
         } else if (fnmatch (GF_XATTR_STIME_PATTERN, key, FNM_NOESCAPE) == 0) {
                 ret = gf_get_min_stime (THIS, dst, key, value);
-                if (ret < 0)
-                        goto out;
+                goto out;
         } else {
                 /* compare user xattrs only */
                 if (!strncmp (key, "user.", strlen ("user."))) {
@@ -160,16 +348,16 @@ dht_aggregate (dict_t *this, char *key, data_t *value, void *data)
                                                       key);
                         }
                 }
-                ret = dict_set (dst, key, value);
-                if (ret) {
-                        gf_msg ("dht", GF_LOG_WARNING, 0,
-                                DHT_MSG_DICT_SET_FAILED,
-                                "Failed to set dictionary value: key = %s",
-                                key);
-                }
         }
 
-        ret = 0;
+        ret = dict_set (dst, key, value);
+        if (ret) {
+                gf_msg ("dht", GF_LOG_WARNING, 0,
+                        DHT_MSG_DICT_SET_FAILED,
+                        "Failed to set dictionary value: key = %s",
+                        key);
+        }
+
 out:
         return ret;
 }
diff --git a/xlators/cluster/dht/src/dht-common.h b/xlators/cluster/dht/src/dht-common.h
index fa06252d80e..29ac798b92b 100644
--- a/xlators/cluster/dht/src/dht-common.h
+++ b/xlators/cluster/dht/src/dht-common.h
@@ -1230,4 +1230,14 @@ dht_get_lock_subvolume (xlator_t *this, struct gf_flock *lock,
 int
 dht_lk_inode_unref (call_frame_t *frame, int32_t op_ret);
 
+int
+add_opt(char **optsp, const char *opt);
+
+char *
+getChoices (const char *value);
+
+int
+dht_aggregate_split_brain_xattr (dict_t *dst, char *key, data_t *value);
+
+
 #endif/* _DHT_H */