[Gluster-users] drop in write performance

Ason Hu tobechihiro at gmail.com
Tue Feb 10 09:01:26 UTC 2015


I got some log from client side


*PARTI*
[2015-02-10 08:23:01.613646] D [MSGID: 0] [dht-common.c:2183:dht_lookup]
0-r2dr3-dht: Calling fresh lookup for
/1w_test00/721ea549-8e2b-4770-9c52-2cc0c4ef872d on r2dr3-replicate-0
[2015-02-10 08:23:01.613947] D [afr-lk-common.c:447:transaction_lk_op]
0-r2dr3-replicate-1: lk op is for a transaction
[2015-02-10 08:23:01.614417] D [MSGID: 0]
[dht-common.c:1819:dht_lookup_cbk] 0-r2dr3-dht: fresh_lookup returned for
/1w_test00/721ea549-8e2b-4770-9c52-2cc0c4ef872d with op_ret -1 and op_errno
2
[2015-02-10 08:23:01.614444] D [MSGID: 0]
[dht-common.c:1823:dht_lookup_cbk] 0-r2dr3-dht: Entry
/1w_test00/721ea549-8e2b-4770-9c52-2cc0c4ef872d missing on subvol
r2dr3-replicate-0
[2015-02-10 08:23:01.614463] D [MSGID: 0]
[dht-common.c:1608:dht_lookup_everywhere] 0-r2dr3-dht: winding lookup call
to 2 subvols
[2015-02-10 08:23:01.615067] D [MSGID: 0]
[dht-common.c:1414:dht_lookup_everywhere_cbk] 0-r2dr3-dht: returned with
op_ret -1 and op_errno 2 (/1w_test00/721ea549-8e2b-4770-9c52-2cc0c4ef872d)
from subvol r2dr3-replicate-1
[2015-02-10 08:23:01.615117] D [MSGID: 0]
[dht-common.c:1414:dht_lookup_everywhere_cbk] 0-r2dr3-dht: returned with
op_ret -1 and op_errno 2 (/1w_test00/721ea549-8e2b-4770-9c52-2cc0c4ef872d)
from subvol r2dr3-replicate-0
[2015-02-10 08:23:01.615139] D [MSGID: 0]
[dht-common.c:1086:dht_lookup_everywhere_done] 0-r2dr3-dht: STATUS:
hashed_subvol r2dr3-replicate-0 cached_subvol null
[2015-02-10 08:23:01.615158] D [MSGID: 0]
[dht-common.c:1147:dht_lookup_everywhere_done] 0-r2dr3-dht: There was no
cached file and  unlink on hashed is not skipped
/1w_test00/721ea549-8e2b-4770-9c52-2cc0c4ef872d
[2015-02-10 08:23:01.615761] D
[afr-transaction.c:1177:afr_post_nonblocking_entrylk_cbk]
0-r2dr3-replicate-0: Non blocking entrylks done. Proceeding to FOP
[2015-02-10 08:23:01.616450] D [afr-lk-common.c:447:transaction_lk_op]
0-r2dr3-replicate-0: lk op is for a transaction
[2015-02-10 08:23:01.617153] D
[afr-transaction.c:1127:afr_post_nonblocking_inodelk_cbk]
0-r2dr3-replicate-0: Non blocking inodelks done. Proceeding to FOP
[2015-02-10 08:23:01.618365] D [MSGID: 0] [dht-common.c:2183:dht_lookup]
0-r2dr3-dht: Calling fresh lookup for
/1w_test00/70e8e730-0442-47c3-94c6-e2d13506427f on r2dr3-replicate-0
[2015-02-10 08:23:01.618670] D [afr-lk-common.c:447:transaction_lk_op]
0-r2dr3-replicate-0: lk op is for a transaction
[2015-02-10 08:23:01.619054] D [MSGID: 0]
[dht-common.c:1819:dht_lookup_cbk] 0-r2dr3-dht: fresh_lookup returned for
/1w_test00/70e8e730-0442-47c3-94c6-e2d13506427f with op_ret -1 and op_errno
2
[2015-02-10 08:23:01.619077] D [MSGID: 0]
[dht-common.c:1823:dht_lookup_cbk] 0-r2dr3-dht: Entry
/1w_test00/70e8e730-0442-47c3-94c6-e2d13506427f missing on subvol
r2dr3-replicate-0
[2015-02-10 08:23:01.619096] D [MSGID: 0]
[dht-common.c:1608:dht_lookup_everywhere] 0-r2dr3-dht: winding lookup call
to 2 subvols
[2015-02-10 08:23:01.619812] D [MSGID: 0]
[dht-common.c:1414:dht_lookup_everywhere_cbk] 0-r2dr3-dht: returned with
op_ret -1 and op_errno 2 (/1w_test00/70e8e730-0442-47c3-94c6-e2d13506427f)
from subvol r2dr3-replicate-1
[2015-02-10 08:23:01.619910] D [MSGID: 0]
[dht-common.c:1414:dht_lookup_everywhere_cbk] 0-r2dr3-dht: returned with
op_ret -1 and op_errno 2 (/1w_test00/70e8e730-0442-47c3-94c6-e2d13506427f)
from subvol r2dr3-replicate-0
[2015-02-10 08:23:01.619953] D [MSGID: 0]
[dht-common.c:1086:dht_lookup_everywhere_done] 0-r2dr3-dht: STATUS:
hashed_subvol r2dr3-replicate-0 cached_subvol null
[2015-02-10 08:23:01.619971] D [MSGID: 0]
[dht-common.c:1147:dht_lookup_everywhere_done] 0-r2dr3-dht: There was no
cached file and  unlink on hashed is not skipped
/1w_test00/70e8e730-0442-47c3-94c6-e2d13506427f
[2015-02-10 08:23:01.620542] D
[afr-transaction.c:1177:afr_post_nonblocking_entrylk_cbk]
0-r2dr3-replicate-0: Non blocking entrylks done. Proceeding to FOP
[2015-02-10 08:23:01.621220] D [afr-lk-common.c:447:transaction_lk_op]
0-r2dr3-replicate-0: lk op is for a transaction
[2015-02-10 08:23:01.621946] D
[afr-transaction.c:1127:afr_post_nonblocking_inodelk_cbk]
0-r2dr3-replicate-0: Non blocking inodelks done. Proceeding to FOP
[2015-02-10 08:23:01.623192] D [MSGID: 0] [dht-common.c:2183:dht_lookup]
0-r2dr3-dht: Calling fresh lookup for
/1w_test00/65dc668b-4b42-45e8-942d-dcb283c206a0 on r2dr3-replicate-1
[2015-02-10 08:23:01.623540] D [afr-lk-common.c:447:transaction_lk_op]
0-r2dr3-replicate-0: lk op is for a transaction
[2015-02-10 08:23:01.623948] D [MSGID: 0]
[dht-common.c:1819:dht_lookup_cbk] 0-r2dr3-dht: fresh_lookup returned for
/1w_test00/65dc668b-4b42-45e8-942d-dcb283c206a0 with op_ret -1 and op_errno
2
[2015-02-10 08:23:01.623975] D [MSGID: 0]
[dht-common.c:1823:dht_lookup_cbk] 0-r2dr3-dht: Entry
/1w_test00/65dc668b-4b42-45e8-942d-dcb283c206a0 missing on subvol
r2dr3-replicate-1
[2015-02-10 08:23:01.623997] D [MSGID: 0]
[dht-common.c:1608:dht_lookup_everywhere] 0-r2dr3-dht: winding lookup call
to 2 subvols
[2015-02-10 08:23:01.624645] D [MSGID: 0]
[dht-common.c:1414:dht_lookup_everywhere_cbk] 0-r2dr3-dht: returned with
op_ret -1 and op_errno 2 (/1w_test00/65dc668b-4b42-45e8-942d-dcb283c206a0)
from subvol r2dr3-replicate-1
[2015-02-10 08:23:01.624746] D [MSGID: 0]
[dht-common.c:1414:dht_lookup_everywhere_cbk] 0-r2dr3-dht: returned with
op_ret -1 and op_errno 2 (/1w_test00/65dc668b-4b42-45e8-942d-dcb283c206a0)
from subvol r2dr3-replicate-0
[2015-02-10 08:23:01.624773] D [MSGID: 0]
[dht-common.c:1086:dht_lookup_everywhere_done] 0-r2dr3-dht: STATUS:
hashed_subvol r2dr3-replicate-1 cached_subvol null
[2015-02-10 08:23:01.624792] D [MSGID: 0]
[dht-common.c:1147:dht_lookup_everywhere_done] 0-r2dr3-dht: There was no
cached file and  unlink on hashed is not skipped
/1w_test00/65dc668b-4b42-45e8-942d-dcb283c206a0
[2015-02-10 08:23:01.625299] D
[afr-transaction.c:1177:afr_post_nonblocking_entrylk_cbk]
0-r2dr3-replicate-1: Non blocking entrylks done. Proceeding to FOP
[2015-02-10 08:23:01.625971] D [afr-lk-common.c:447:transaction_lk_op]
0-r2dr3-replicate-1: lk op is for a transaction
[2015-02-10 08:23:01.626690] D
[afr-transaction.c:1127:afr_post_nonblocking_inodelk_cbk]
0-r2dr3-replicate-1: Non blocking inodelks done. Proceeding to FOP


*PARTII*
[2015-02-10 08:48:30.396145] D [rpc-clnt-ping.c:255:rpc_clnt_start_ping]
0-r2dr3-client-1: returning as transport is already disconnected OR there
are no frames (0 || 0)
[2015-02-10 08:48:37.396576] D [rpc-clnt-ping.c:255:rpc_clnt_start_ping]
0-r2dr3-client-3: returning as transport is already disconnected OR there
are no frames (0 || 0)
[2015-02-10 08:48:38.396684] D [rpc-clnt-ping.c:255:rpc_clnt_start_ping]
0-r2dr3-client-2: returning as transport is already disconnected OR there
are no frames (0 || 0)
[2015-02-10 08:49:13.398692] D [rpc-clnt-ping.c:255:rpc_clnt_start_ping]
0-r2dr3-client-0: returning as transport is already disconnected OR there
are no frames (0 || 0)
[2015-02-10 08:49:13.398751] D [rpc-clnt-ping.c:255:rpc_clnt_start_ping]
0-r2dr3-client-1: returning as transport is already disconnected OR there
are no frames (0 || 0)
[2015-02-10 08:49:20.399164] D [rpc-clnt-ping.c:255:rpc_clnt_start_ping]
0-r2dr3-client-3: returning as transport is already disconnected OR there
are no frames (0 || 0)


*Volume INFO*
Volume Name: r2dr3
Type: Distributed-Replicate
Volume ID: b30ffc1a-9a6c-45e7-a847-97efd9d864d2
Status: Started
Number of Bricks: 2 x 2 = 4
Transport-type: tcp
Bricks:
Brick1: node1:/gfs_data/r2dr3
Brick2: node2:/gfs_data/r2dr3
Brick3: node3:/gfs_data/r2dr3
Brick4: node4:/gfs_data/r2dr3
Options Reconfigured:
diagnostics.count-fop-hits: on
diagnostics.latency-measurement: on
diagnostics.brick-log-level: DEBUG
diagnostics.client-log-level: DEBUG


2015-02-06 18:32 GMT+08:00 Ason Hu <tobechihiro at gmail.com>:

> Hi Pranith,
>
>
> Gluster Version :
> glusterfs 3.6.2 built on Jan 22 2015 12:58:10
>
>    - 100,000 file test in XFS directly
>
> [root at glusterfs-node1 gfs_data]# java FileCreationAndSearchtest 10w_test
> 100000
> Creation finished!! Time: *8961(ms)*
> File Search Finished!! 33223 searches, time: 124ms, misses:0
> NIO Search Finished!! 33223 searches, time: 367ms, misses:0
>
>
>
>    - 100,000 file test from client (Distributed Replicated)
>
> [root at glusterfs-client01 drvol]# java FileCreationAndSearchtest
> 10w_test01 100000
> Start!!
> Creation finished!! Time: *265713(ms)*
> File Search Finished!! 33453 searches, time: 18403(ms), misses:0
> NIO Search Finished!! 33453 searches, time: 17827(ms), misses:0
>
>
>
> after many small file created (≒200 million), it cost more than
> 1000000(ms) to finish 100,000 file test
>
> I've flushed all volumes and create new one to reproduce the condition
> again.
>
>
>
> 2015-02-05 19:19 GMT+08:00 Pranith Kumar Karampuri <pkarampu at redhat.com>:
>
>>
>> On 02/02/2015 02:42 PM, Ason Hu wrote:
>>
>> Hi,
>>
>>  In my small file test, 3 types of volume are created in same GlusterFS
>> Cluster (4 nodes)
>>
>>    - Distritubed
>>     - Distributed Replicated
>>     - Distributed Striped
>>
>>
>> 10 million small files (8 byte) create and search each test cycle life
>>
>>
>>  Nodes info :
>>
>>    - OS : CentOS 6.6 64bit
>>    - RAM : 64GB
>>     - HDD : 4TB, 7200RPM, SAS *6 RAID 0
>>     - FileSystem : XFS, BS=4K
>>
>>
>>
>>  Problem :
>>
>>  It occur when file reach ≒200 million (sum of 3 volumes)
>>  files created speed drop from  ≒1500 to  ≒100
>>
>> If you do this test directly on the XFS partition, what is the kind of
>> performance difference you see?
>>
>> Pranith
>>
>>
>>  Though the speed of creation become slow but search speed and
>> throughput still good
>> Seems each types of volume have same problem, I don't know what cause
>> this problem..
>>
>>
>>  Workaround :
>>
>>  I try to create new volume ( Distributed | Distributed Replicated |
>> Distributed Striped ), mount it,
>> problem solved!
>> But the old volume write speed still sad..
>>
>>
>>
>>  I'd  try to increase below volume setting, but don't work at all
>>
>>    - performance.cache-size
>>     - performance.io-thread-count
>>
>>
>>  I didn't enable the volume qouta option.
>> Does glusterfs volume have some kind of limitation of file creation?
>>
>>
>>
>>  Regards
>> Nosa Hu,
>>
>>
>>
>>
>> _______________________________________________
>> Gluster-users mailing listGluster-users at gluster.orghttp://www.gluster.org/mailman/listinfo/gluster-users
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150210/e422ccba/attachment.html>


More information about the Gluster-users mailing list