[jira] [Created] (HBASE-14676) HBaseTestCase clean out: Purge Incommon Interface and Table and Region implementations
stack created HBASE-14676: - Summary: HBaseTestCase clean out: Purge Incommon Interface and Table and Region implementations Key: HBASE-14676 URL: https://issues.apache.org/jira/browse/HBASE-14676 Project: HBase Issue Type: Bug Reporter: stack As part of the hollowing out of the old HBaseTestCase in preparation for removal, this patch purges the old Incommon trick that made it so you could pass an Interface to a method to do loading or testing and the implementation could be an HTable or a Region. Instead, replace it with a RegionTable, a Regoin that has the Table Interface as a decoration so you can use same loading code and same test both on top of the Region or on other side of network via HTable. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Created] (HBASE-14675) Excise deprecated Put#add(...) and replace with Put#addColumn(...)
Jonathan Hsieh created HBASE-14675: -- Summary: Excise deprecated Put#add(...) and replace with Put#addColumn(...) Key: HBASE-14675 URL: https://issues.apache.org/jira/browse/HBASE-14675 Project: HBase Issue Type: Improvement Affects Versions: 2.0.0 Reporter: Jonathan Hsieh Assignee: Jonathan Hsieh Fix For: 2.0.0 The Put API changed from #add(...) to #addColumn(...). This updates all instances of it and removes it from the Put (which was added for hbase 1.0.0) -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Could all tests run MR without MiniMapReduceCluster?
When i dig in HBASE-14659, i found in testcase TestHFileOutputFormat, some tests run MR job without MiniMRCluster and some run with it. The ones with no MiniMRCluster run much faster! Do we need to 'test' clustered MR especially in a unit test? As stack suggestion, i post this topic here for discussion.
[jira] [Created] (HBASE-14677) BucketAllocator freeBlock is the bottleneck in bulk block cache evictions
Vladimir Rodionov created HBASE-14677: - Summary: BucketAllocator freeBlock is the bottleneck in bulk block cache evictions Key: HBASE-14677 URL: https://issues.apache.org/jira/browse/HBASE-14677 Project: HBase Issue Type: Bug Components: BlockCache Affects Versions: 0.98.15, 1.0.1.1, 1.1.1, 1.0.2 Reporter: Vladimir Rodionov BucketCache implementation of freeBlock has a bad performance / concurrency and affects bulk block evictions from cache on a file close (after split, compactions or region moves). -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Resolved] (HBASE-14677) BucketAllocator freeBlock is the bottleneck in bulk block cache evictions
[ https://issues.apache.org/jira/browse/HBASE-14677?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Vladimir Rodionov resolved HBASE-14677. --- Resolution: Duplicate Duplicate of HBASE-14624 > BucketAllocator freeBlock is the bottleneck in bulk block cache evictions > - > > Key: HBASE-14677 > URL: https://issues.apache.org/jira/browse/HBASE-14677 > Project: HBase > Issue Type: Bug > Components: BlockCache >Affects Versions: 1.0.2, 1.1.1, 1.0.1.1, 0.98.15 >Reporter: Vladimir Rodionov > > BucketCache implementation of freeBlock has a bad performance / concurrency > and affects bulk block evictions from cache on a file close (after split, > compactions or region moves). -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Created] (HBASE-14678) Experiment: Temporarily disable balancer and a few others to see if root of crashed/timedout JVMs
stack created HBASE-14678: - Summary: Experiment: Temporarily disable balancer and a few others to see if root of crashed/timedout JVMs Key: HBASE-14678 URL: https://issues.apache.org/jira/browse/HBASE-14678 Project: HBase Issue Type: Sub-task Reporter: stack Looking at recent builds of 1.2, I see a few of the runs finishing with kills and notice that a JVM exited without reporting back state. Running the hanging test finder, I can see at least that in one case that the balancer tests seem to be outstanding; looking in test output, seems to be still going on A few others are reported as hung but they look like they have just started running and are just killed by surefire. This issue is about trying to disable a few of the problematics like balancer tests to see if our overall stability improves. If so, I can concentrate on stabilizing these few tests. Else will just undo the experiment and put the tests back on line. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Re: Could all tests run MR without MiniMapReduceCluster?
Do we have any tests that test the locality feature for map task -> machine? That's the only thing I can think of that would 100% require an mr mincluster. I'd be +1 on removing most of the uses of mr minicluster. On Wed, Oct 21, 2015 at 10:01 PM, 陈恒wrote: > When i dig in HBASE-14659, i found in testcase TestHFileOutputFormat, > some tests run MR job > without MiniMRCluster and some run with it. > > > The ones with no MiniMRCluster run much faster! > Do we need to 'test' clustered MR especially in a unit test? > > > As stack suggestion, i post this topic here for discussion.
Re: Could all tests run MR without MiniMapReduceCluster?
On Wed, Oct 21, 2015 at 10:01 PM, 陈恒wrote: > When i dig in HBASE-14659, i found in testcase TestHFileOutputFormat, > some tests run MR job > without MiniMRCluster and some run with it. > > > The ones with no MiniMRCluster run much faster! > Do we need to 'test' clustered MR especially in a unit test? > > > We have some fat unit tests that spin up loads of mini clusters. For MR, the same code base is exercised whether cluster or not so if no MR cluster means less resources used by test and all runs faster, I'd be +1 in favor of undoing cluster (for all but the cases where cluster needed to prove some code as per example Elliott raises above). Thanks 陈恒 St.Ack
Re: add msg in disable table interface
Isn't such information already available in master log (though not on one line) ? bq. a process need to add cp so disable Using online table alter feature, there is no need to disable the whole table. Cheers On Thu, Oct 22, 2015 at 3:07 AM, wangyongqiang0...@163.com < wangyongqiang0...@163.com> wrote: > we can find some tables are disabled sometimes, and we want to chcek out > why. > when we check master log, we can see logs are as follows: > Client=root//"ip_address disable" "table_name" > > this log is from hbase0.98.10, line 2213 > > can we add a parameter named message to the interface disableTable, we > can add some explain-msg, such as > hbase-shell disable table(we can know this table is disabled by hand) > a process need to add cp so disable > a process need to drop table so disable > > we can more easily know why this table is disabled > > > > wangyongqiang0...@163.com >
[jira] [Created] (HBASE-14679) Separate security shell tests from other shell tests
Ted Yu created HBASE-14679: -- Summary: Separate security shell tests from other shell tests Key: HBASE-14679 URL: https://issues.apache.org/jira/browse/HBASE-14679 Project: HBase Issue Type: Task Reporter: Ted Yu Assignee: Ted Yu Priority: Minor TestShell takes a long time to run, sometimes timing out (especially on branch-1). This issue proposes the following changes: separate security shell tests into their own test class use unique table names for different tests -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Created] (HBASE-14681) Upgrade Checkstyle plugin to 2.15 to avoid MCHECKSTYLE-288
Misty Stanley-Jones created HBASE-14681: --- Summary: Upgrade Checkstyle plugin to 2.15 to avoid MCHECKSTYLE-288 Key: HBASE-14681 URL: https://issues.apache.org/jira/browse/HBASE-14681 Project: HBase Issue Type: Bug Components: build Affects Versions: 2.0.0 Reporter: Misty Stanley-Jones Assignee: Misty Stanley-Jones Fix For: 2.0.0 We are getting a NPE in checkstyle when running mvn:site. It seems to be MCHECKSTYLE-288. Updating maven-checkstyle-plugin from 2.13 to 2.15 seems to fix it. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Created] (HBASE-14682) CM restore functionality for regionservers is broken
Enis Soztutar created HBASE-14682: - Summary: CM restore functionality for regionservers is broken Key: HBASE-14682 URL: https://issues.apache.org/jira/browse/HBASE-14682 Project: HBase Issue Type: Bug Reporter: Enis Soztutar Assignee: Enis Soztutar In the {{DistributedHBaseCluster.restoreClusterStatus()}}, we are starting a regionserver for backup masters. Seems to be a copy-paste error from HBASE-12429. This is causing further issues on our test rig since we are starting a regionserver where only a backup master were before: {code} INFO [main] hbase.HBaseClusterManager: Executing remote command: ps aux | grep proc_master | grep -v grep | tr -s ' ' | cut -d ' ' -f2 , hostname:10.0.0.99 INFO [main] util.Shell: Executing full command [/usr/bin/ssh -o StrictHostKeyChecking=no 10.0.0.99 "sudo su - hbase -c \"ps aux | grep proc_master | grep -v grep | tr -s ' ' | cut -d ' ' -f2\""] INFO [main] hbase.HBaseClusterManager: Executed remote command, exit code:0 , output:13244 INFO [main] hbase.HBaseClusterManager: Executing remote command: ps aux | grep proc_regionserver | grep -v grep | tr -s ' ' | cut -d ' ' -f2 , hostname:10.0.0.99 INFO [main] util.Shell: Executing full command [/usr/bin/ssh -o StrictHostKeyChecking=no 10.0.0.99 "sudo su - hbase -c \"ps aux | grep proc_regionserver | grep -v grep | tr -s ' ' | cut -d ' ' -f2\""] INFO [main] hbase.HBaseClusterManager: Executed remote command, exit code:0 , output: INFO [main] hbase.HBaseCluster: Restoring cluster - starting initial region server: 10.0.0.99:16000 INFO [main] hbase.HBaseCluster: Starting RS on: 10.0.0.99 INFO [main] hbase.HBaseClusterManager: Executing remote command: /usr/hdp/2.3.3.0-2971/hbase/bin/hbase-daemon.sh --config /tmp/hbaseConf start regionserver , hostname:10.0.0.99 INFO [main] util.Shell: Executing full command [/usr/bin/ssh -o StrictHostKeyChecking=no 10.0.0.99 "sudo su - hbase -c \"/usr/hdp/2.3.3.0-2971/hbase/bin/hbase-daemon.sh --config /tmp/hbaseConf start reg INFO [main] hbase.HBaseClusterManager: Executed remote command, exit code:0 , output:starting regionserver, logging to /var/log/hbase/hbase-hbase-regionserver-zk0-hb23-h.out INFO [main] hbase.HBaseClusterManager: Executing remote command: ps aux | grep proc_regionserver | grep -v grep | tr -s ' ' | cut -d ' ' -f2 , hostname:10.0.0.126 INFO [main] util.Shell: Executing full command [/usr/bin/ssh -o StrictHostKeyChecking=no 10.0.0.126 "sudo su - hbase -c \"ps aux | grep proc_regionserver | grep -v grep | tr -s ' ' | cut -d ' ' -f2\""] INFO [main] hbase.HBaseClusterManager: Executed remote command, exit code:0 , output: INFO [main] hbase.HBaseCluster: Added new HBaseAdmin INFO [main] hbase.HBaseCluster: Restoring cluster - done DEBUG [main] hbase.IntegrationTestBase: Done restoring the cluster {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Resolved] (HBASE-14681) Upgrade Checkstyle plugin to 2.16
[ https://issues.apache.org/jira/browse/HBASE-14681?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Misty Stanley-Jones resolved HBASE-14681. - Resolution: Fixed Hadoop Flags: Reviewed > Upgrade Checkstyle plugin to 2.16 > - > > Key: HBASE-14681 > URL: https://issues.apache.org/jira/browse/HBASE-14681 > Project: HBase > Issue Type: Bug > Components: build >Affects Versions: 2.0.0 >Reporter: Misty Stanley-Jones >Assignee: Misty Stanley-Jones > Fix For: 2.0.0 > > Attachments: 2.13_fail.log, 2.15_success.log.zip, HBASE-14681.patch > > > We are getting a NPE in checkstyle when running mvn:site. It seems to be > MCHECKSTYLE-288 or MCHECKSTYLE-250. Updating maven-checkstyle-plugin from > 2.13 to 2.16 seems to fix it. > Or maybe not. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
add msg in disable table interface
we can find some tables are disabled sometimes, and we want to chcek out why. when we check master log, we can see logs are as follows: Client=root//"ip_address disable" "table_name" this log is from hbase0.98.10, line 2213 can we add a parameter named message to the interface disableTable, we can add some explain-msg, such as hbase-shell disable table(we can know this table is disabled by hand) a process need to add cp so disable a process need to drop table so disable we can more easily know why this table is disabled wangyongqiang0...@163.com
Upgrading Apache HTrace integration to 4.x branch
Hey guys, I am wondering if there are there any plans on upgrading Apache HTrace version used in HBase client / server from 3.x to 4.x release branch? Unfortunately, HTrace 3.x and 4.x versions are not compatible with respect to integration modules (zipkin, htraced) and cause nasty classpath conflicts. Thanks! Best Regards, Andriy Redko
[jira] [Created] (HBASE-14683) Batching in buffered mutator is awful when adding lists of mutations.
Elliott Clark created HBASE-14683: - Summary: Batching in buffered mutator is awful when adding lists of mutations. Key: HBASE-14683 URL: https://issues.apache.org/jira/browse/HBASE-14683 Project: HBase Issue Type: Bug Reporter: Elliott Clark Assignee: Elliott Clark -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Re: Could all tests run MR without MiniMapReduceCluster?
On Thu, Oct 22, 2015 at 7:41 PM, Heng Chenwrote: > IMO we can remove them firstly, and see what QA says. > +1
Opinions wanted: new site skin
Hi all, We are currently using the reFlow Maven site skin. I went looking around and found Fluido, which seems to be a bit more extensible. I built and staged a version of the site at http://mstanleyjones.github.io/hbase/index.html. Note the Github ribbon and the Google site search. I'm curious to know what you think. I also put the 0.94 docs menu as a submenu of the Documentation menu, to see how it looked. Thanks, Misty
Re: Could all tests run MR without MiniMapReduceCluster?
What Elliott said. St.Ack On Thu, Oct 22, 2015 at 10:09 PM, Elliott Clarkwrote: > On Thu, Oct 22, 2015 at 7:41 PM, Heng Chen > wrote: > > > IMO we can remove them firstly, and see what QA says. > > > > +1 >
Re: Opinions wanted: new site skin
I think it looks pretty amazing. Fresh! I like the menu of menu reports. It shines light on mess we've shoved under the carpet for ages now (dependencies, being public w/ what the licensing of our depenencies are...). Can we change the google search to be search-hadoop.com setup for hbase searches? +1 St.Ack On Thu, Oct 22, 2015 at 10:17 PM, Misty Stanley-Jones < mstanleyjo...@cloudera.com> wrote: > Hi all, > > We are currently using the reFlow Maven site skin. I went looking around > and found Fluido, which seems to be a bit more extensible. I built and > staged a version of the site at > http://mstanleyjones.github.io/hbase/index.html. Note the Github ribbon > and > the Google site search. I'm curious to know what you think. > > I also put the 0.94 docs menu as a submenu of the Documentation menu, to > see how it looked. > > Thanks, > Misty >
Re: Upgrading Apache HTrace integration to 4.x branch
Looks like there's work to do that happening at https://issues.apache.org/jira/browse/HBASE-14451 On Thu, Oct 22, 2015 at 7:47 AM, Andrey Redkowrote: > Hey guys, > > I am wondering if there are there any plans on upgrading Apache HTrace > version used in HBase client / server from 3.x to 4.x release branch? > Unfortunately, HTrace 3.x and 4.x versions are not compatible with respect > to integration modules (zipkin, htraced) and cause nasty classpath > conflicts. > > Thanks! > > Best Regards, > Andriy Redko >
Re: Upgrading Apache HTrace integration to 4.x branch
What Dave said. The posted patch will put us up on htrace 4. Lacking is the work to make it so traces tell a good story, that all the previous trace points are properly connected and perhaps adding more. Should get to it soon. Thanks for asking, St.Ack On Thu, Oct 22, 2015 at 8:49 AM, Dave Lathamwrote: > Looks like there's work to do that happening at > https://issues.apache.org/jira/browse/HBASE-14451 > > On Thu, Oct 22, 2015 at 7:47 AM, Andrey Redko wrote: > > > Hey guys, > > > > I am wondering if there are there any plans on upgrading Apache HTrace > > version used in HBase client / server from 3.x to 4.x release branch? > > Unfortunately, HTrace 3.x and 4.x versions are not compatible with > respect > > to integration modules (zipkin, htraced) and cause nasty classpath > > conflicts. > > > > Thanks! > > > > Best Regards, > > Andriy Redko > > >