Bug 1144792

Summary: Very high memory usage during rebalance
Product: [Community] GlusterFS Reporter: Krutika Dhananjay <kdhananj>
Component: distributeAssignee: Krutika Dhananjay <kdhananj>
Status: CLOSED CURRENTRELEASE QA Contact:
Severity: medium Docs Contact:
Priority: unspecified    
Version: 3.4.5CC: bugs, kdhananj, kkeithle, rgowdapp, ryan.clough
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: 1142052 Environment:
Last Closed: 2015-04-13 07:10:10 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 985957, 1125245    

Description Krutika Dhananjay 2014-09-21 02:05:54 UTC 
+++ This bug was initially created as a clone of Bug #1142052 +++

Description of problem:
Rebalance has been running for about 7 days on a 2 node, 2 brick, 52TB distributed volume. Memory usage has grown slowing over time to consume all available physical memory. OOM killer stopped the last rebalance and then after restarting all Gluster processes I am attempting it again.
"sync; echo 3 > /proc/sys/vm/drop_caches" has no effect to lower memory consumption.

Version-Release number of selected component (if applicable):
[root@hgluster01 ~]# glusterfs --version
glusterfs 3.5.2 built on Jul 31 2014 18:47:52

How reproducible:
So far, I cannot complete a rebalance.

Steps to Reproduce:
1. Start rebalance
gluster volume rebalance export_volume start

Actual results:
High memory consumption by glusterfs process gets OOM'd

Expected results:
Rebalance does not consume all available memory and completes the rebalance and fix-layout.

Additional info:
[root@hgluster01 ~]# gluster volume status export_volume detail
Status of volume: export_volume
------------------------------------------------------------------------------
Brick                : Brick hgluster01:/gluster_data
Port                 : 49152               
Online               : Y                   
Pid                  : 2438                
File System          : xfs                 
Device               : /dev/mapper/vg_data-lv_data
Mount Options        : rw,noatime,nodiratime,logbufs=8,logbsize=256k,inode64,nobarrier
Inode Size           : 512                 
Disk Space Free      : 12.3TB              
Total Disk Space     : 27.3TB              
Inode Count          : 2929685696          
Free Inodes          : 2839872616          
------------------------------------------------------------------------------
Brick                : Brick hgluster02:/gluster_data
Port                 : 49152               
Online               : Y                   
Pid                  : 2467                
File System          : xfs                 
Device               : /dev/mapper/vg_data-lv_data
Mount Options        : rw,noatime,nodiratime,logbufs=8,logbsize=256k,inode64,nobarrier
Inode Size           : 512                 
Disk Space Free      : 12.4TB              
Total Disk Space     : 27.3TB              
Inode Count          : 2929685696          
Free Inodes          : 2839847441

[root@hgluster01 ~]# gluster volume info
 
Volume Name: export_volume
Type: Distribute
Volume ID: c74cc970-31e2-4924-a244-4c70d958dadb
Status: Started
Number of Bricks: 2
Transport-type: tcp
Bricks:
Brick1: hgluster01:/gluster_data
Brick2: hgluster02:/gluster_data
Options Reconfigured:
performance.stat-prefetch: on
performance.write-behind: on
performance.flush-behind: on
features.quota-deem-statfs: on
performance.quick-read: on
performance.client-io-threads: on
performance.read-ahead: on
performance.io-thread-count: 24
features.quota: on
cluster.eager-lock: on
nfs.disable: on
auth.allow: 192.168.10.*,10.0.10.*,10.8.0.*
server.allow-insecure: on
performance.write-behind-window-size: 4MB
network.ping-timeout: 60
features.quota-timeout: 10
performance.io-cache: off

[root@hgluster01 ~]# cat /proc/meminfo 
MemTotal:       32844100 kB
MemFree:         2148772 kB
Buffers:           14184 kB
Cached:            35600 kB
SwapCached:       204288 kB
Active:         24682388 kB
Inactive:        3315448 kB
Active(anon):   24660896 kB
Inactive(anon):  3289292 kB
Active(file):      21492 kB
Inactive(file):    26156 kB
Unevictable:       12728 kB
Mlocked:            4552 kB
SwapTotal:      16490488 kB
SwapFree:       15077012 kB
Dirty:                32 kB
Writeback:             0 kB
AnonPages:      27761596 kB
Mapped:             9168 kB
Shmem:                 4 kB
Slab:             544552 kB
SReclaimable:     273636 kB
SUnreclaim:       270916 kB
KernelStack:        4800 kB
PageTables:        60592 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:    32912536 kB
Committed_AS:   29529576 kB
VmallocTotal:   34359738367 kB
VmallocUsed:      345236 kB
VmallocChunk:   34340927412 kB
HardwareCorrupted:     0 kB
AnonHugePages:  17307648 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
DirectMap4k:        4728 kB
DirectMap2M:     2058240 kB
DirectMap1G:    31457280 kB

[root@hgluster01 ~]# pmap -x 2627
2627:   /usr/sbin/glusterfs -s localhost --volfile-id export_volume --xlator-option *dht.use-readdirp=yes --xlator-option *dht.lookup-unhashed=yes --xlator-option *dht.assert-no-child-down=yes --xlator-option *replicate*.data-self-heal=off --xlator-option *replicate*.metadata-self-heal=off --xlator-option *replicate*.entry-self-heal=off --xlator-option *replicate*.readdir-failover=off --xlator-option *dht.readdir-optimize=on --xlator-option *dht.rebalance-cmd=1 --xlator-option *dht.node-uuid=875dbae1-82bd-485f-98e
Address           Kbytes     RSS   Dirty Mode   Mapping
0000000000400000      64      20       0 r-x--  glusterfsd
0000000000610000       8       8       8 rw---  glusterfsd
0000000001a15000     132      40      40 rw---    [ anon ]
0000000001a36000 28914064 27616220 27611240 rw---    [ anon ]
00007f0290000000     132      16      16 rw---    [ anon ]
00007f0290021000   65404       0       0 -----    [ anon ]
00007f0297a1e000    6024    1184    1184 rw---    [ anon ]
00007f0298000000     132       0       0 rw---    [ anon ]
00007f0298021000   65404       0       0 -----    [ anon ]
00007f029c000000     132      28      28 rw---    [ anon ]
00007f029c021000   65404       0       0 -----    [ anon ]
00007f02a0000000     132       8       8 rw---    [ anon ]
00007f02a0021000   65404       0       0 -----    [ anon ]
00007f02a4092000    4964      16      16 rw---    [ anon ]
00007f02a456b000       4       0       0 -----    [ anon ]
00007f02a456c000   11844    2056    2056 rw---    [ anon ]
00007f02a50fd000       4       0       0 -----    [ anon ]
00007f02a50fe000    1024       8       8 rw---    [ anon ]
00007f02a51fe000      96      12       0 r-x--  io-stats.so
00007f02a5216000    2048       0       0 -----  io-stats.so
00007f02a5416000       8       0       0 rw---  io-stats.so
00007f02a5418000      96      16       0 r-x--  io-threads.so
00007f02a5430000    2044       0       0 -----  io-threads.so
00007f02a562f000      12       4       4 rw---  io-threads.so
00007f02a5632000      52       0       0 r-x--  md-cache.so
00007f02a563f000    2044       0       0 -----  md-cache.so
00007f02a583e000       8       0       0 rw---  md-cache.so
00007f02a5840000      28       0       0 r-x--  open-behind.so
00007f02a5847000    2048       0       0 -----  open-behind.so
00007f02a5a47000       4       0       0 rw---  open-behind.so
00007f02a5a48000      28       0       0 r-x--  quick-read.so
00007f02a5a4f000    2044       0       0 -----  quick-read.so
00007f02a5c4e000       8       0       0 rw---  quick-read.so
00007f02a5c50000      44       0       0 r-x--  read-ahead.so
00007f02a5c5b000    2044       0       0 -----  read-ahead.so
00007f02a5e5a000       8       0       0 rw---  read-ahead.so
00007f02a5e5c000      48       0       0 r-x--  write-behind.so
00007f02a5e68000    2048       0       0 -----  write-behind.so
00007f02a6068000       8       0       0 rw---  write-behind.so
00007f02a606a000     300     136       0 r-x--  dht.so
00007f02a60b5000    2048       0       0 -----  dht.so
00007f02a62b5000      16       8       8 rw---  dht.so
00007f02a62b9000     240     108       0 r-x--  client.so
00007f02a62f5000    2048       0       0 -----  client.so
00007f02a64f5000      16       8       8 rw---  client.so
00007f02a64f9000       4       0       0 -----    [ anon ]
00007f02a64fa000   10240       8       8 rw---    [ anon ]
00007f02a6efa000      48      12       0 r-x--  libnss_files-2.12.so
00007f02a6f06000    2048       0       0 -----  libnss_files-2.12.so
00007f02a7106000       4       0       0 r----  libnss_files-2.12.so
00007f02a7107000       4       0       0 rw---  libnss_files-2.12.so
00007f02a7108000     116       0       0 r-x--  libselinux.so.1
00007f02a7125000    2044       0       0 -----  libselinux.so.1
00007f02a7324000       4       0       0 r----  libselinux.so.1
00007f02a7325000       4       0       0 rw---  libselinux.so.1
00007f02a7326000       4       0       0 rw---    [ anon ]
00007f02a7327000      88       0       0 r-x--  libresolv-2.12.so
00007f02a733d000    2048       0       0 -----  libresolv-2.12.so
00007f02a753d000       4       0       0 r----  libresolv-2.12.so
00007f02a753e000       4       0       0 rw---  libresolv-2.12.so
00007f02a753f000       8       0       0 rw---    [ anon ]
00007f02a7541000       8       0       0 r-x--  libkeyutils.so.1.3
00007f02a7543000    2044       0       0 -----  libkeyutils.so.1.3
00007f02a7742000       4       0       0 r----  libkeyutils.so.1.3
00007f02a7743000       4       0       0 rw---  libkeyutils.so.1.3
00007f02a7744000      40       0       0 r-x--  libkrb5support.so.0.1
00007f02a774e000    2044       0       0 -----  libkrb5support.so.0.1
00007f02a794d000       4       0       0 r----  libkrb5support.so.0.1
00007f02a794e000       4       0       0 rw---  libkrb5support.so.0.1
00007f02a794f000     164       0       0 r-x--  libk5crypto.so.3.1
00007f02a7978000    2048       0       0 -----  libk5crypto.so.3.1
00007f02a7b78000       4       0       0 r----  libk5crypto.so.3.1
00007f02a7b79000       4       0       0 rw---  libk5crypto.so.3.1
00007f02a7b7a000       4       0       0 rw---    [ anon ]
00007f02a7b7b000      12       0       0 r-x--  libcom_err.so.2.1
00007f02a7b7e000    2044       0       0 -----  libcom_err.so.2.1
00007f02a7d7d000       4       0       0 r----  libcom_err.so.2.1
00007f02a7d7e000       4       0       0 rw---  libcom_err.so.2.1
00007f02a7d7f000     876       0       0 r-x--  libkrb5.so.3.3
00007f02a7e5a000    2044       0       0 -----  libkrb5.so.3.3
00007f02a8059000      40       4       4 r----  libkrb5.so.3.3
00007f02a8063000       8       0       0 rw---  libkrb5.so.3.3
00007f02a8065000     260       0       0 r-x--  libgssapi_krb5.so.2.2
00007f02a80a6000    2048       0       0 -----  libgssapi_krb5.so.2.2
00007f02a82a6000       4       0       0 r----  libgssapi_krb5.so.2.2
00007f02a82a7000       8       4       4 rw---  libgssapi_krb5.so.2.2
00007f02a82a9000     388       4       0 r-x--  libssl.so.1.0.1e
00007f02a830a000    2048       0       0 -----  libssl.so.1.0.1e
00007f02a850a000      16       0       0 r----  libssl.so.1.0.1e
00007f02a850e000      28       0       0 rw---  libssl.so.1.0.1e
00007f02a8515000      60      44       0 r-x--  socket.so
00007f02a8524000    2048       0       0 -----  socket.so
00007f02a8724000      16       8       8 rw---  socket.so
00007f02a8728000       4       0       0 -----    [ anon ]
00007f02a8729000   10240       8       8 rw---    [ anon ]
00007f02a9129000       4       0       0 -----    [ anon ]
00007f02a912a000   10240       8       8 rw---    [ anon ]
00007f02a9b2a000       4       0       0 -----    [ anon ]
00007f02a9b2b000   10240       0       0 rw---    [ anon ]
00007f02aa52b000   20052    6296    6296 rw---    [ anon ]
00007f02ab8c0000      84       0       0 r-x--  libz.so.1.2.3
00007f02ab8d5000    2044       0       0 -----  libz.so.1.2.3
00007f02abad4000       4       0       0 r----  libz.so.1.2.3
00007f02abad5000       4       0       0 rw---  libz.so.1.2.3
00007f02abad6000    1576     756       0 r-x--  libc-2.12.so
00007f02abc60000    2048       0       0 -----  libc-2.12.so
00007f02abe60000      16      16      16 r----  libc-2.12.so
00007f02abe64000       4       4       4 rw---  libc-2.12.so
00007f02abe65000      20      16      16 rw---    [ anon ]
00007f02abe6a000    1748       0       0 r-x--  libcrypto.so.1.0.1e
00007f02ac01f000    2048       0       0 -----  libcrypto.so.1.0.1e
00007f02ac21f000     108       0       0 r----  libcrypto.so.1.0.1e
00007f02ac23a000      48       0       0 rw---  libcrypto.so.1.0.1e
00007f02ac246000      16       0       0 rw---    [ anon ]
00007f02ac24a000      92      72       0 r-x--  libpthread-2.12.so
00007f02ac261000    2048       0       0 -----  libpthread-2.12.so
00007f02ac461000       4       4       4 r----  libpthread-2.12.so
00007f02ac462000       4       4       4 rw---  libpthread-2.12.so
00007f02ac463000      16       4       4 rw---    [ anon ]
00007f02ac467000      28       4       0 r-x--  librt-2.12.so
00007f02ac46e000    2044       0       0 -----  librt-2.12.so
00007f02ac66d000       4       4       4 r----  librt-2.12.so
00007f02ac66e000       4       0       0 rw---  librt-2.12.so
00007f02ac66f000    1396       0       0 r-x--  libpython2.6.so.1.0
00007f02ac7cc000    2044       0       0 -----  libpython2.6.so.1.0
00007f02ac9cb000     240       0       0 rw---  libpython2.6.so.1.0
00007f02aca07000      56       0       0 rw---    [ anon ]
00007f02aca15000     524       0       0 r-x--  libm-2.12.so
00007f02aca98000    2044       0       0 -----  libm-2.12.so
00007f02acc97000       4       0       0 r----  libm-2.12.so
00007f02acc98000       4       0       0 rw---  libm-2.12.so
00007f02acc99000       8       0       0 r-x--  libutil-2.12.so
00007f02acc9b000    2044       0       0 -----  libutil-2.12.so
00007f02ace9a000       4       0       0 r----  libutil-2.12.so
00007f02ace9b000       4       0       0 rw---  libutil-2.12.so
00007f02ace9c000       8       0       0 r-x--  libdl-2.12.so
00007f02ace9e000    2048       0       0 -----  libdl-2.12.so
00007f02ad09e000       4       0       0 r----  libdl-2.12.so
00007f02ad09f000       4       0       0 rw---  libdl-2.12.so
00007f02ad0a0000      88      24       0 r-x--  libgfxdr.so.0.0.0
00007f02ad0b6000    2044       0       0 -----  libgfxdr.so.0.0.0
00007f02ad2b5000       4       4       4 rw---  libgfxdr.so.0.0.0
00007f02ad2b6000      96      64       0 r-x--  libgfrpc.so.0.0.0
00007f02ad2ce000    2048       0       0 -----  libgfrpc.so.0.0.0
00007f02ad4ce000       4       4       4 rw---  libgfrpc.so.0.0.0
00007f02ad4cf000     532     176       0 r-x--  libglusterfs.so.0.0.0
00007f02ad554000    2048       0       0 -----  libglusterfs.so.0.0.0
00007f02ad754000       8       8       8 rw---  libglusterfs.so.0.0.0
00007f02ad756000      16      12      12 rw---    [ anon ]
00007f02ad75a000     128      96       0 r-x--  ld-2.12.so
00007f02ad7ac000    1824      24      24 rw---    [ anon ]
00007f02ad977000       4       0       0 rw---    [ anon ]
00007f02ad978000       4       0       0 rw---    [ anon ]
00007f02ad979000       4       4       4 r----  ld-2.12.so
00007f02ad97a000       4       0       0 rw---  ld-2.12.so
00007f02ad97b000       4       0       0 rw---    [ anon ]
00007fff4c597000     124      96      96 rw---    [ stack ]
00007fff4c5ff000       4       4       0 r-x--    [ anon ]
ffffffffff600000       4       0       0 r-x--    [ anon ]
----------------  ------  ------  ------
total kB        29338940 27627692 27621164

--- Additional comment from Raghavendra G on 2014-09-18 02:53:33 EDT ---

Hi Ryan,

If rebalance is still running, can you please get the statedump of rebalance process on all the nodes which are part of volume? 

We have to repeat following steps on all the nodes which are part of volume.

1. Get the pid of the rebalance process:

[root@unused glusterfs]# ps ax | grep -i rebalance  | grep glusterfs | cut -d" " -f 1
16537

2. Get the statedump of rebalance process:
[root@unused glusterfs]# kill -SIGUSR1 16537

3. statedump can be found in /var/run/gluster/

[root@unused glusterfs]# ls /var/run/gluster/*16537*
/var/run/gluster/glusterdump.16537.dump.1411022946


regards,
Raghavendra.

--- Additional comment from Anand Avati on 2014-09-18 05:40:52 EDT ---

REVIEW: http://review.gluster.org/8763 (cluster/dht: Fix dict_t leaks in rebalance process' execution path) posted (#1) for review on master by Krutika Dhananjay (kdhananj)

--- Additional comment from Krutika Dhananjay on 2014-09-18 06:39:40 EDT ---

Hi Ryan,

Thanks for the bug report. We have identified few leaks in rebalance and are in the process of fixing them.

--- Additional comment from Ryan Clough on 2014-09-18 13:05:11 EDT ---

Unfortunately, the processes had been OOM'd before I got to the office this morning. The reblance failed.

--- Additional comment from Anand Avati on 2014-09-19 03:26:41 EDT ---

REVIEW: http://review.gluster.org/8776 (cluster/dht: Fix dict_t leaks in rebalance process' execution path) posted (#1) for review on release-3.5 by Krutika Dhananjay (kdhananj)

--- Additional comment from Anand Avati on 2014-09-19 04:13:24 EDT ---

REVIEW: http://review.gluster.org/8763 (cluster/dht: Fix dict_t leaks in rebalance process' execution path) posted (#2) for review on master by Krutika Dhananjay (kdhananj)

--- Additional comment from Anand Avati on 2014-09-19 10:10:33 EDT ---

COMMIT: http://review.gluster.org/8763 committed in master by Vijay Bellur (vbellur) 
------
commit 258e61adb5505124925c71d2a0d0375d086e32d4
Author: Krutika Dhananjay <kdhananj>
Date:   Thu Sep 18 14:36:38 2014 +0530

    cluster/dht: Fix dict_t leaks in rebalance process' execution path
    
    Two dict_t objects are leaked for every file migrated in success codepath.
    It is the caller's responsibility to unref dict that it gets from calls to
    syncop_getxattr(); and rebalance performs two syncop_getxattr()s per file
    without freeing them.
    
    Also, syncop_getxattr() on GF_XATTR_LINKINFO_KEY doesn't seem to be using
    the response dict. Hence, NULL is now passed as opposed to @dict to
    syncop_getxattr().
    
    Change-Id: I5a4b5ab834df3633dea994f239bbdbc34cbe9259
    BUG: 1142052
    Signed-off-by: Krutika Dhananjay <kdhananj>
    Reviewed-on: http://review.gluster.org/8763
    Tested-by: Gluster Build System <jenkins.com>
    Reviewed-by: Shyamsundar Ranganathan <srangana>
    Reviewed-by: Vijay Bellur <vbellur>

--- Additional comment from Anand Avati on 2014-09-19 23:44:45 EDT ---

REVIEW: http://review.gluster.org/8784 (cluster/dht: Fix dict_t leaks in rebalance process' execution path) posted (#1) for review on release-3.5 by Krutika Dhananjay (kdhananj)
Comment 1 Anand Avati 2014-09-21 02:18:46 UTC 
REVIEW: http://review.gluster.org/8789 (cluster/dht: Fix dict_t leaks in rebalance process' execution path) posted (#1) for review on release-3.4 by Krutika Dhananjay (kdhananj)
Comment 2 Anand Avati 2014-09-23 11:03:17 UTC 
COMMIT: http://review.gluster.org/8789 committed in release-3.4 by Kaleb KEITHLEY (kkeithle) 
------
commit e2a76e76e3d9174fa77f40f3abd0212705283f00
Author: Krutika Dhananjay <kdhananj>
Date:   Thu Sep 18 14:36:38 2014 +0530

    cluster/dht: Fix dict_t leaks in rebalance process' execution path
    
            Backport of: http://review.gluster.org/8763
    
    Two dict_t objects are leaked for every file migrated in success codepath.
    It is the caller's responsibility to unref dict that it gets from calls to
    syncop_getxattr(); and rebalance performs two syncop_getxattr()s per file
    without freeing them.
    
    Also, syncop_getxattr() on GF_XATTR_LINKINFO_KEY doesn't seem to be using
    the response dict. Hence, NULL is now passed as opposed to @dict to
    syncop_getxattr().
    
    Change-Id: I59556ee6e135e7e65d4ddd31ba0f39e7eb50b02d
    BUG: 1144792
    Signed-off-by: Krutika Dhananjay <kdhananj>
    Reviewed-on: http://review.gluster.org/8789
    Tested-by: Gluster Build System <jenkins.com>
    Reviewed-by: Kaleb KEITHLEY <kkeithle>
Comment 3 Kaleb KEITHLEY 2014-10-20 15:10:14 UTC 
gluster workflow is

  POST: patch posted to gerrit
  MODIFIED: patch merged to source tree
Comment 4 Kaleb KEITHLEY 2014-10-20 15:11:10 UTC 
but I've already merged it, so don't mind me.