From e589d8de66d3325da8fbbbe44d1a5bd6335e08ab Mon Sep 17 00:00:00 2001
From: Mohit Agrawal <moagrawal@redhat.com>
Date: Wed, 4 Mar 2020 09:17:26 +0530
Subject: core[brick_mux]: brick crashed when creating and deleting volumes
 over time

Problem: In brick_mux environment, while volumes are created/stopped in a loop
         after running a long time the main brick is crashed.The brick is crashed
         because the main brick process was not cleaned up memory for all objects
         at the time of detaching a volume.
         Below are the objects that are missed at the time of detaching a volume
         1) xlator object for a brick graph
         2) local_pool for posix_lock xlator
         3) rpc object cleanup at quota xlator
         4) inode leak at brick xlator

Solution: To avoid the crash resolve all leak at the time of detaching a brick
Change-Id: Ibb6e46c5fba22b9441a88cbaf6b3278823235913
updates: #977
Signed-off-by: Mohit Agrawal <moagrawa@redhat.com>
---
 libglusterfs/src/xlator.c | 34 ++++++++++++++++++++++++++--------
 1 file changed, 26 insertions(+), 8 deletions(-)

(limited to 'libglusterfs')

diff --git a/libglusterfs/src/xlator.c b/libglusterfs/src/xlator.c
index 90a993afdd2..e793e1e3d68 100644
--- a/libglusterfs/src/xlator.c
+++ b/libglusterfs/src/xlator.c
@@ -937,6 +937,8 @@ xlator_mem_cleanup(xlator_t *this)
     xlator_list_t **trav_p = NULL;
     xlator_t *top = NULL;
     xlator_t *victim = NULL;
+    glusterfs_graph_t *graph = NULL;
+    gf_boolean_t graph_cleanup = _gf_false;
 
     if (this->call_cleanup || !this->ctx)
         return;
@@ -944,6 +946,12 @@ xlator_mem_cleanup(xlator_t *this)
     this->call_cleanup = 1;
     ctx = this->ctx;
 
+    inode_table = this->itable;
+    if (inode_table) {
+        inode_table_destroy(inode_table);
+        this->itable = NULL;
+    }
+
     xlator_call_fini(trav);
 
     while (prev) {
@@ -952,12 +960,6 @@ xlator_mem_cleanup(xlator_t *this)
         prev = trav;
     }
 
-    inode_table = this->itable;
-    if (inode_table) {
-        inode_table_destroy(inode_table);
-        this->itable = NULL;
-    }
-
     if (this->fini) {
         this->fini(this);
     }
@@ -967,17 +969,28 @@ xlator_mem_cleanup(xlator_t *this)
     if (ctx->active) {
         top = ctx->active->first;
         LOCK(&ctx->volfile_lock);
-        /* TODO here we have leak for xlator node in a graph */
-        /* Need to move only top xlator from a graph */
         for (trav_p = &top->children; *trav_p; trav_p = &(*trav_p)->next) {
             victim = (*trav_p)->xlator;
             if (victim->call_cleanup && !strcmp(victim->name, this->name)) {
+                graph_cleanup = _gf_true;
                 (*trav_p) = (*trav_p)->next;
                 break;
             }
         }
         UNLOCK(&ctx->volfile_lock);
     }
+
+    if (graph_cleanup) {
+        prev = this;
+        graph = ctx->active;
+        pthread_mutex_lock(&graph->mutex);
+        while (prev) {
+            trav = prev->next;
+            GF_FREE(prev);
+            prev = trav;
+        }
+        pthread_mutex_unlock(&graph->mutex);
+    }
 }
 
 void
@@ -1544,6 +1557,11 @@ graph_total_client_xlator(glusterfs_graph_t *graph)
     }
 
     xl = graph->first;
+    if (!strcmp(xl->type, "protocol/server")) {
+        gf_msg_debug(xl->name, 0, "Return because it is a server graph");
+        return 0;
+    }
+
     while (xl) {
         if (strcmp(xl->type, "protocol/client") == 0) {
             count++;
-- 
cgit