hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bing Jiang (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-8801) Region level read/write degradation
Date Tue, 02 Jul 2013 08:02:23 GMT

    [ https://issues.apache.org/jira/browse/HBASE-8801?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13697596#comment-13697596
] 

Bing Jiang commented on HBASE-8801:
-----------------------------------

 it is an interesting issue.I think the read/write degradation should contains:

1.Based on table/region, it needs provide the interface to configure table, besides region
alone.

2.It is not enough to switch off/on region read/write, and whether to provide the read/write
throughput level degradation or not.

In my opinion,if set tag in .META., it will make more costs from maintenance, and not friendly
with hbase-version upgrade. 

If this degradation is just designed to be an administrator's tool, maybe org.apache.hadoop.hbase.WritableRpcEngine
is a very good choice.
When set 'org.apache.hadoop.ipc.HBaseServer.trace' DEBUG, it will see:
{quote}
2013-07-02 15:43:54,160 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #327005; Served:
HRegionInterface#get queueTime=0 processingTime=2 contents=1 Get, 85 bytes}
2013-07-02 15:43:54,331 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #195052; Served:
HRegionInterface#multi queueTime=0 processingTime=1 contents=5 Puts, 105 values [ min=0 max=416
avg=29 ], 105 KeyValues
2013-07-02 15:43:54,529 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #327009; Served:
HRegionInterface#get queueTime=0 processingTime=341 contents=1 Get, 86 bytes
2013-07-02 15:43:54,532 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #327010; Served:
HRegionInterface#get queueTime=0 processingTime=1 contents=1 Get, 85 bytes
2013-07-02 15:43:56,537 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #327094; Served:
HRegionInterface#multi queueTime=0 processingTime=1 contents=4 Puts, 98 values [ min=0 max=297
avg=19 ], 98 KeyValues
2013-07-02 15:43:57,053 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #182856; Served:
HRegionInterface#multi queueTime=0 processingTime=0 contents=3 Puts, 15 values [ min=1 max=16
avg=9 ], 15 KeyValues
processingTime=2 contents=6 Puts, 168 values [ min=0 max=119 avg=11 ], 168 KeyValues
2013-07-02 15:43:57,179 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #176328; Served:
HRegionInterface#multi queueTime=0 processingTime=0 contents=8 Puts, 40 values [ min=1 max=16
avg=8 ], 40 KeyValues
2013-07-02 15:43:57,249 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #327109; Served:
HRegionInterface#get queueTime=0 processingTime=1 contents=1 Get, 85 bytes
2013-07-02 15:43:57,341 DEBUG org.apache.hadoop.ipc.HBaseServer.trace: Call #509143; Served:
HRegionInterface#multi queueTime=1 processingTime=0 contents=5 Puts, 140 values [ min=0 max=119
avg=11 ], 140 KeyValues
processingTime=0 contents=2 Puts, 10 values [ min=1 max=16 avg=9 ], 10 KeyValues
{quote}
And try to add table/region filter in WritableRpcEngine, will function in an easier way, or
maybe it can be used on interface(multi/put/get/openScanner/..) to control the region or table's
throughput.
:)


                
> Region level read/write degradation
> -----------------------------------
>
>                 Key: HBASE-8801
>                 URL: https://issues.apache.org/jira/browse/HBASE-8801
>             Project: HBase
>          Issue Type: New Feature
>            Reporter: Chao Shi
>
> We would like to propose a tool for HBase administrators to disable read and/or write
for single region temporarily.
> Our HBase table at weibo.com is accessed by many clients. We have experienced the following
outage: due to bugs of user code, a small number of regions are reading at very high rate.
Requests to other regions at the same RSs are affected. In such scenario, the admin would
like to temporarily set the region to write-only (only forbids reads so writes can still being
served).

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message