[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13180392#comment-13180392
]
ramkrishna.s.vasudevan commented on HBASE-4899:
---
@Stack and @Chunhui
This defect is not present in 0.90 branch? But the fixed version says it is in
0.90?
Do we need to add in 0.90.6?
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Fix For: 0.90.0
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13181152#comment-13181152
]
ramkrishna.s.vasudevan commented on HBASE-4899:
---
Updated as committed only to Trunk and 0.92 and not in 0.90
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Fix For: 0.92.0
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13162081#comment-13162081
]
Hudson commented on HBASE-4899:
---
Integrated in HBase-TRUNK-security #19 (See
[https://builds.apache.org/job/HBase-TRUNK-security/19/])
HBASE-4899 Region would be assigned twice easily with continually killing
server and moving region in testing environment
stack :
Files :
*
/hbase/trunk/src/main/java/org/apache/hadoop/hbase/master/handler/ServerShutdownHandler.java
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Fix For: 0.92.1
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13161473#comment-13161473
]
Hudson commented on HBASE-4899:
---
Integrated in HBase-0.92 #168 (See
[https://builds.apache.org/job/HBase-0.92/168/])
HBASE-4899 Region would be assigned twice easily with continually killing
server and moving region in testing environment
stack :
Files :
* /hbase/branches/0.92/CHANGES.txt
*
/hbase/branches/0.92/src/main/java/org/apache/hadoop/hbase/master/handler/ServerShutdownHandler.java
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Fix For: 0.92.1
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13161514#comment-13161514
]
Hudson commented on HBASE-4899:
---
Integrated in HBase-TRUNK #2508 (See
[https://builds.apache.org/job/HBase-TRUNK/2508/])
HBASE-4899 Region would be assigned twice easily with continually killing
server and moving region in testing environment
stack :
Files :
*
/hbase/trunk/src/main/java/org/apache/hadoop/hbase/master/handler/ServerShutdownHandler.java
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Fix For: 0.92.1
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13162059#comment-13162059
]
Hudson commented on HBASE-4899:
---
Integrated in HBase-0.92-security #29 (See
[https://builds.apache.org/job/HBase-0.92-security/29/])
HBASE-4899 Region would be assigned twice easily with continually killing
server and moving region in testing environment
stack :
Files :
* /hbase/branches/0.92/CHANGES.txt
*
/hbase/branches/0.92/src/main/java/org/apache/hadoop/hbase/master/handler/ServerShutdownHandler.java
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Fix For: 0.92.1
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13159921#comment-13159921
]
chunhui shen commented on HBASE-4899:
-
@Ted
{code}
+ + because it has been opened in
+ + addressFromAM.getServerName());
{code}
To the above log, the region must has been opend, because the return of
AssignmentManager#getRegionServerOfRegion(region) is other server.
Test is running now.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Assignee: chunhui shen
Attachments: hbase-4899.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160018#comment-13160018
]
chunhui shen commented on HBASE-4899:
-
{code}
Results :
Tests run: 1176, Failures: 0, Errors: 0, Skipped: 9
[INFO]
[INFO] --- maven-surefire-plugin:2.11-TRUNK-HBASE-2:test
(secondPartTestsExecution) @ hbase ---
[INFO] Tests are skipped.
[INFO]
[INFO] BUILD SUCCESS
[INFO]
[INFO] Total time: 1:43:30.281s
[INFO] Finished at: Wed Nov 30 20:20:20 CST 2011
[INFO] Final Memory: 35M/361M
[INFO]
{code}
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Assignee: chunhui shen
Attachments: hbase-4899.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160025#comment-13160025
]
chunhui shen commented on HBASE-4899:
-
In patchv1,{code}continue;{code} is forgot to add.
The above is test results with patchV2.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Assignee: chunhui shen
Attachments: hbase-4899.patch, hbase-4899v2.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160057#comment-13160057
]
Hadoop QA commented on HBASE-4899:
--
-1 overall. Here are the results of testing the latest attachment
http://issues.apache.org/jira/secure/attachment/12505620/hbase-4899v2.patch
against trunk revision .
+1 @author. The patch does not contain any @author tags.
-1 tests included. The patch doesn't appear to include any new or modified
tests.
Please justify why no new tests are needed for this
patch.
Also please list what manual steps were performed to
verify this patch.
-1 javadoc. The javadoc tool appears to have generated -162 warning
messages.
+1 javac. The applied patch does not increase the total number of javac
compiler warnings.
-1 findbugs. The patch appears to introduce 67 new Findbugs (version
1.3.9) warnings.
+1 release audit. The applied patch does not increase the total number of
release audit warnings.
-1 core tests. The patch failed these unit tests:
org.apache.hadoop.hbase.mapreduce.TestTimeRangeMapRed
org.apache.hadoop.hbase.mapreduce.TestLoadIncrementalHFilesSplitRecovery
org.apache.hadoop.hbase.TestFullLogReconstruction
org.apache.hadoop.hbase.mapreduce.TestImportTsv
org.apache.hadoop.hbase.mapreduce.TestTableMapReduce
Test results:
https://builds.apache.org/job/PreCommit-HBASE-Build/407//testReport/
Findbugs warnings:
https://builds.apache.org/job/PreCommit-HBASE-Build/407//artifact/trunk/patchprocess/newPatchFindbugsWarnings.html
Console output: https://builds.apache.org/job/PreCommit-HBASE-Build/407//console
This message is automatically generated.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Assignee: chunhui shen
Attachments: hbase-4899.patch, hbase-4899v2.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160437#comment-13160437
]
stack commented on HBASE-4899:
--
Good find. Here are some comments on the patch.
{code}
+if(rit!=null!rit.isClosing() !rit.isPendingClose()){
{code}
Our style is to put spaces between operators as in 'if (rit != null ...' etc.
+ LOG.debug(Skip assgining region
The above should be assigning.
+ + rit.getRegion().getRegionNameAsString()
+ + because in RIT region state: + rit.getState());
I think just to rit.toString()?
+ continue;
This is confusing. Is this code inside a loop and this continue takes us to
the head of the loop again? Its buried pretty deeply here so I'd not be
surprised if readers missed it. Shouldn't you rather write the code as if/else?
+if (addressFromAM != null
!addressFromAM.equals(this.serverName)) {
+ LOG.debug(Skip assgining region
Again, a misspelling.
I can fix the above small things on commit or maybe you want to have a go at it?
This is a pretty nice find. Thanks for the patch.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Attachments: hbase-4899.patch, hbase-4899v2.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160574#comment-13160574
]
chunhui shen commented on HBASE-4899:
-
@stack
I has amend it in patchV3.
Thanks.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160611#comment-13160611
]
Hadoop QA commented on HBASE-4899:
--
-1 overall. Here are the results of testing the latest attachment
http://issues.apache.org/jira/secure/attachment/12505704/hbase-4899v3.patch
against trunk revision .
+1 @author. The patch does not contain any @author tags.
-1 tests included. The patch doesn't appear to include any new or modified
tests.
Please justify why no new tests are needed for this
patch.
Also please list what manual steps were performed to
verify this patch.
-1 javadoc. The javadoc tool appears to have generated -160 warning
messages.
+1 javac. The applied patch does not increase the total number of javac
compiler warnings.
-1 findbugs. The patch appears to introduce 71 new Findbugs (version
1.3.9) warnings.
+1 release audit. The applied patch does not increase the total number of
release audit warnings.
-1 core tests. The patch failed these unit tests:
org.apache.hadoop.hbase.master.TestRollingRestart
org.apache.hadoop.hbase.master.TestRestartCluster
org.apache.hadoop.hbase.thrift2.TestThriftHBaseServiceHandler
org.apache.hadoop.hbase.master.TestDistributedLogSplitting
org.apache.hadoop.hbase.master.TestHMasterRPCException
org.apache.hadoop.hbase.mapreduce.TestTimeRangeMapRed
org.apache.hadoop.hbase.master.TestMaster
org.apache.hadoop.hbase.mapreduce.TestLoadIncrementalHFilesSplitRecovery
org.apache.hadoop.hbase.master.TestMasterRestartAfterDisablingTable
org.apache.hadoop.hbase.TestDrainingServer
org.apache.hadoop.hbase.mapreduce.TestTableInputFormatScan
org.apache.hadoop.hbase.master.TestZKBasedOpenCloseRegion
org.apache.hadoop.hbase.TestFullLogReconstruction
org.apache.hadoop.hbase.avro.TestAvroServer
org.apache.hadoop.hbase.mapreduce.TestLoadIncrementalHFiles
org.apache.hadoop.hbase.master.TestMasterFailover
org.apache.hadoop.hbase.mapreduce.TestImportTsv
org.apache.hadoop.hbase.master.TestMasterTransitions
org.apache.hadoop.hbase.mapreduce.TestHFileOutputFormat
org.apache.hadoop.hbase.master.TestSplitLogManager
org.apache.hadoop.hbase.master.TestOpenedRegionHandler
Test results:
https://builds.apache.org/job/PreCommit-HBASE-Build/413//testReport/
Findbugs warnings:
https://builds.apache.org/job/PreCommit-HBASE-Build/413//artifact/trunk/patchprocess/newPatchFindbugsWarnings.html
Console output: https://builds.apache.org/job/PreCommit-HBASE-Build/413//console
This message is automatically generated.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160630#comment-13160630
]
stack commented on HBASE-4899:
--
+1 on patch.
Running tests locally before commit. The above failures shouldn't be because
of this patch.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13160677#comment-13160677
]
chunhui shen commented on HBASE-4899:
-
Testing result on my QA environment
{code}
Results :
Tests run: 1175, Failures: 0, Errors: 0, Skipped: 9
[INFO]
[INFO] --- maven-surefire-plugin:2.11-TRUNK-HBASE-2:test
(secondPartTestsExecution) @ hbase ---
[INFO] Tests are skipped.
[INFO]
[INFO] BUILD SUCCESS
[INFO]
[INFO] Total time: 1:44:10.984s
[INFO] Finished at: Thu Dec 01 14:10:34 CST 2011
[INFO] Final Memory: 35M/380M
[INFO]
{code}
please check!
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.1
Reporter: chunhui shen
Assignee: chunhui shen
Priority: Critical
Attachments: hbase-4899.patch, hbase-4899v2.patch, hbase-4899v3.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13159858#comment-13159858
]
Ted Yu commented on HBASE-4899:
---
{code}
+ + because it has been opened in
+ + addressFromAM.getServerName());
{code}
We should use the value of rit (RegionState) in the above log instead of hard
coding 'opened'.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Attachments: hbase-4899.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13159859#comment-13159859
]
Ted Yu commented on HBASE-4899:
---
@Chunhui:
Please let us know testing result on your QA environment.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Assignee: chunhui shen
Attachments: hbase-4899.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4899) Region would be assigned twice easily with continually killing server and moving region in testing environment
[
https://issues.apache.org/jira/browse/HBASE-4899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13159881#comment-13159881
]
Hadoop QA commented on HBASE-4899:
--
-1 overall. Here are the results of testing the latest attachment
http://issues.apache.org/jira/secure/attachment/12505580/hbase-4899.patch
against trunk revision .
+1 @author. The patch does not contain any @author tags.
-1 tests included. The patch doesn't appear to include any new or modified
tests.
Please justify why no new tests are needed for this
patch.
Also please list what manual steps were performed to
verify this patch.
-1 javadoc. The javadoc tool appears to have generated -162 warning
messages.
+1 javac. The applied patch does not increase the total number of javac
compiler warnings.
-1 findbugs. The patch appears to introduce 67 new Findbugs (version
1.3.9) warnings.
+1 release audit. The applied patch does not increase the total number of
release audit warnings.
-1 core tests. The patch failed these unit tests:
org.apache.hadoop.hbase.TestFullLogReconstruction
org.apache.hadoop.hbase.mapreduce.TestImportTsv
Test results:
https://builds.apache.org/job/PreCommit-HBASE-Build/406//testReport/
Findbugs warnings:
https://builds.apache.org/job/PreCommit-HBASE-Build/406//artifact/trunk/patchprocess/newPatchFindbugsWarnings.html
Console output: https://builds.apache.org/job/PreCommit-HBASE-Build/406//console
This message is automatically generated.
Region would be assigned twice easily with continually killing server and
moving region in testing environment
---
Key: HBASE-4899
URL: https://issues.apache.org/jira/browse/HBASE-4899
Project: HBase
Issue Type: Bug
Affects Versions: 0.92.0
Reporter: chunhui shen
Assignee: chunhui shen
Attachments: hbase-4899.patch
Before assigning region in ServerShutdownHandler#process, it will check
whether region is in RIT,
however, this checking doesn't work as the excepted in the following case:
1.move region A from server B to server C
2.kill server B
3.start server B immediately
Let's see what happen in the code for the above case
{code}
for step1:
1.1 server B close the region A,
1.2 master setOffline for region
A,(AssignmentManager#setOffline:this.regions.remove(regionInfo))
1.3 server C start to open region A.(Not completed)
for step3:
master ServerShutdownHandler#process() for server B
{
..
splitlog()
...
ListRegionState regionsInTransition =
this.services.getAssignmentManager()
.processServerShutdown(this.serverName);
...
Skip regions that were in transition unless CLOSING or PENDING_CLOSE
...
assign region
}
{code}
In fact, when running
ServerShutdownHandler#process()#this.services.getAssignmentManager().processServerShutdown(this.serverName),
region A is in RIT (step1.3 not completed), but the return ListRegionState
regionsInTransition doesn't contain it, because region A has removed from
AssignmentManager.regions by AssignmentManager#setOffline in step 1.2
Therefore, region A will be assigned twice.
Actually, one server killed and started twice will also easily cause region
assigned twice.
Exclude the above reason, another probability :
when execute ServerShutdownHandler#process()#MetaReader.getServerUserRegions
,region is included which is in RIT now.
But after completing MetaReader.getServerUserRegions, the region has been
opened in other server and is not in RIT now.
In our testing environment where balancing,moving and killing are executed
periodly, assigning region twice often happens, and it is hateful because it
will affect other test cases.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
