bq. it would have been better to have included at that time the detailed
explanation
Will pay attention next time such situation arises.
Cheers
On Thu, Dec 19, 2013 at 1:47 PM, Andrew Purtell wrote:
> I reviewed this change. As I understand, neither the code before and after
> is incorrect, b
I reviewed this change. As I understand, neither the code before and after
is incorrect, but the improved behavior after the change allows us to pass
reliably on Hadoop 2 with its timing changes. I did fail to ask Ted to
write up a release note for the JIRA, so I am sorry for that.
Also, Ted, if y
Vote passes with 5 binding +1's. ( apurtell, eclark, stack, enis, jmhsieh).
Pushing out the release and announcement to user@.
Thanks,
Jon.
On Thu, Dec 19, 2013 at 12:12 PM, Stack wrote:
> On Thu, Dec 19, 2013 at 8:39 AM, Jonathan Hsieh wrote:
>
> > ...
> >
> > I've spent some time reviewing
On Thu, Dec 19, 2013 at 8:39 AM, Jonathan Hsieh wrote:
> ...
>
> I've spent some time reviewing HBASE-10142, There are some non-test code
> modifications still trying to determine if it is a serious problem or not
> on that side. Ted, is there a reason why this wasn't ported to the 0.96
> branc
Thanks Ted.
I'm looked at the history of that code and believe that this race and test
case only causes a perf problem (we may roll extra times) and has been
around since 0.90.5.
I'm +1'ing this for release also.
On Thu, Dec 19, 2013 at 9:45 AM, Ted Yu wrote:
> There used to be some comment a
There used to be some comment around low replication checking:
// TODO: preserving the old behavior for now, but this check is
strange. It's not // protected by any locks here, so for
all we know rolling locks might start // as soon as we
enter the "if". Is this best-ef
Ted,
My question really boils down this this -- can we have a data loss if we
don't take in HBASE-10142 or not? It will take me a little more time
reverse engineer and convince myself one way or another.If we can lose
data, then I'll roll take the port and do another rc. If we don't (we are
Jon:
If Stack gives the greenlight, I can certainly port it to 0.96 branch.
Cheers
On Thu, Dec 19, 2013 at 8:39 AM, Jonathan Hsieh wrote:
> When I run the test standalone, and didn't have an failure in 0.96.1.1rc0
> or 0.96.1. When I ran the whole suite, I ran into exactly the same failure
>
When I run the test standalone, and didn't have an failure in 0.96.1.1rc0
or 0.96.1. When I ran the whole suite, I ran into exactly the same failure
on 0.96.1.1 (currnetly testing full suite from 0.96.1 src tar ball)
I've spent some time reviewing HBASE-10142, There are some non-test code
modific
The test failure you mentioned has been fixed by HBASE-10142.
Cheers
On Wed, Dec 18, 2013 at 9:00 PM, Jean-Marc Spaggiari <
jean-m...@spaggiari.org> wrote:
> Typo is because I have no done a cut&past ;)
>
> With cut&past: mvn test -PrunAllTests -Dsurefire.secondPartThreadCount=8
>
> On the last
Typo is because I have no done a cut&past ;)
With cut&past: mvn test -PrunAllTests -Dsurefire.secondPartThreadCount=8
On the last run, error is:
Failed tests:
testLogRollOnDatanodeDeath(org.apache.hadoop.hbase.regionserver.wal.TestLogRolling):
LowReplication Roller should've been disabled, curren
What error(s) did you see ?
There was a typo in this def ('s' between D and u):
-Dsurefire.secondPartThreadCount
You can lower the thread count: in trunk build, value of 2 is used.
Cheers
On Wed, Dec 18, 2013 at 8:45 PM, Jean-Marc Spaggiari <
jean-m...@spaggiari.org> wrote:
> I tried multiple
I tried multiple times over many hours to run:
mvn test -PrunallTests -Dusefire.secondPartThreadCount=8
On a local machine using the src jar, with no success. I might be missing
something... I will investigate so I will be able to provide better
feedback for 0.96.2...
Sorry about that.
2013/12/
+1.
- downloaded the artifacts
- checked checksums
- checked sigs
- checked hadoop libs in h1 / h2
- checked directory layouts
- run local cluster
- run smoke tests with shell on the artifacts
- run tests locally:
-- bin/hbase org.apache.hadoop.hbase.util.LoadTestTool -write 10:10:100
-num_keys
+1
Downloaded, unbundled, checked layout, and ran it. Browsed the mvn
artifacts.
St.Ack
On Tue, Dec 17, 2013 at 2:23 PM, Jonathan Hsieh wrote:
> This is a quick-fix release directly off of the 0.96.1 release. It can be
> downloaded here:
>
> http://people.apache.org/~jmhsieh/hbase-0.96.1.1
+1 Check signature
Unpacked tar
Checked structure.
ran rat check.
On Tue, Dec 17, 2013 at 4:51 PM, Andrew Purtell wrote:
> +1
>
> Checked signatures and file sums
> Unpacked tarballs - Contents and directory structure look good
> Started up miniclusters out of unpacked directories
>
>
>
>
> On
+1
Checked signatures and file sums
Unpacked tarballs - Contents and directory structure look good
Started up miniclusters out of unpacked directories
On Tue, Dec 17, 2013 at 2:23 PM, Jonathan Hsieh wrote:
> This is a quick-fix release directly off of the 0.96.1 release. It can be
> downlo
This is a quick-fix release directly off of the 0.96.1 release. It can be
downloaded here:
http://people.apache.org/~jmhsieh/hbase-0.96.1.1rc0/
The maven staging repo is here:
https://repository.apache.org/content/repositories/orgapachehbase-056/
There is only one jira'ed patch in this releas
18 matches
Mail list logo