You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by Mikhail Antonov <ol...@gmail.com> on 2016/10/29 00:17:46 UTC
State of 1.3 first RC
Guys,
I've been working on getting the first RC out this week and want to share
an update on the current state of the things.
- HBASE-16951 - was found and fixed, when we don't include
hbase-archetypes in the tarball assembly
- HBASE-16570 - optimization that caused a regression in the balancer, it
was reverted from 1.3. Once fixed that would go to 1.4 and perhaps 1.3.1.
- HBASE-16743 - flaky TestSimpleRpcScheduler, fixed now (I ran several
cycles of 50 iterations each and all have passed).
Now, there's presently one remaining blocker - HBASE-16964, when we don't
properly handle archiving of store files after compactions (thanks Gary for
digging into that!).
See the jira for details.
Also, as a note - the above is a third major bug around compactions that we
saw on 1.3 during our testing last month (see also HBASE-16788
and HBASE-16754)..
So during the upcoming RC testing let's try to stress this area.
Until 1.3 is out let's consider branch-1.3 to be in a lockdown for new
changes. Ping me if you think there is something that I'm missing that
should go out in 1.3. Otherwise,
important but non-blocker fixes in flight would go in 1.3.1.
Stay tuned!
Mikhail
Re: State of 1.3 first RC
Posted by Yu Li <ca...@gmail.com>.
Have committed HBASE-16960 today, FYI @Mikhail
Looking forward to the 1st RC (Smile)
Best Regards,
Yu
On 4 November 2016 at 06:09, Mikhail Antonov <ol...@gmail.com> wrote:
> All rights guys, thanks for the pings and patience! Status update:
>
> - all patches listed in the first email on the thread have been committed
> - in addition to those following fixes will go in (good catches!)
> * HBASE-16931 Setting cell's seqId to zero in compaction flow might
> cause RS down (committed)
> * HBASE-16960 RegionServer hang when aborting (committed to master and
> branch-1, pending commit to branch-1.3).
>
> With that I'm planning to prepare the RC. The other patches pinged upon
> would be in 1.3.1 or 1.4.0.
> Speak up if I missed some pending fixes you consider critical.
>
> Thanks!
> -Mikhail
>
> On Tue, Nov 1, 2016 at 10:46 AM, Mikhail Antonov <ol...@gmail.com>
> wrote:
>
> > Looking - thanks!
> >
> > -M
> >
> > On Tue, Nov 1, 2016 at 10:03 AM, Yu Li <ca...@gmail.com> wrote:
> >
> >> Another critical one for your attention sir (patch ready for commit):
> >> https://issues.apache.org/jira/browse/HBASE-16960
> >>
> >> Best Regards,
> >> Yu
> >>
> >> On 31 October 2016 at 16:59, Mikhail Antonov <ol...@gmail.com>
> >> wrote:
> >>
> >> > Hi Yu,
> >> >
> >> > sorry I missed that and thanks for reaching out. Let me have a look
> now
> >> and
> >> > comment on the jira.
> >> >
> >> > -Mikhail
> >> >
> >> > On Mon, Oct 31, 2016 at 12:30 AM, Yu Li <ca...@gmail.com> wrote:
> >> >
> >> > > I think HBASE-16931 is also a critical issue (oops also in
> compaction
> >> > > area...) and should go in 1.3.0, pinged several times there but
> >> failed to
> >> > > draw your attention boss (smile). Mind take a look? Thanks.
> >> > >
> >> > > Btw, is there any ETA for 1.3.0 yet?
> >> > >
> >> > > Best Regards,
> >> > > Yu
> >> > >
> >> > > On 29 October 2016 at 08:17, Mikhail Antonov <ol...@gmail.com>
> >> > wrote:
> >> > >
> >> > > > Guys,
> >> > > >
> >> > > > I've been working on getting the first RC out this week and want
> to
> >> > share
> >> > > > an update on the current state of the things.
> >> > > >
> >> > > > - HBASE-16951 - was found and fixed, when we don't include
> >> > > > hbase-archetypes in the tarball assembly
> >> > > > - HBASE-16570 - optimization that caused a regression in the
> >> balancer,
> >> > > it
> >> > > > was reverted from 1.3. Once fixed that would go to 1.4 and perhaps
> >> > 1.3.1.
> >> > > > - HBASE-16743 - flaky TestSimpleRpcScheduler, fixed now (I ran
> >> several
> >> > > > cycles of 50 iterations each and all have passed).
> >> > > >
> >> > > > Now, there's presently one remaining blocker - HBASE-16964, when
> we
> >> > don't
> >> > > > properly handle archiving of store files after compactions (thanks
> >> Gary
> >> > > for
> >> > > > digging into that!).
> >> > > > See the jira for details.
> >> > > >
> >> > > > Also, as a note - the above is a third major bug around
> compactions
> >> > that
> >> > > we
> >> > > > saw on 1.3 during our testing last month (see also HBASE-16788
> >> > > > and HBASE-16754)..
> >> > > > So during the upcoming RC testing let's try to stress this area.
> >> > > >
> >> > > > Until 1.3 is out let's consider branch-1.3 to be in a lockdown for
> >> new
> >> > > > changes. Ping me if you think there is something that I'm missing
> >> that
> >> > > > should go out in 1.3. Otherwise,
> >> > > > important but non-blocker fixes in flight would go in 1.3.1.
> >> > > >
> >> > > > Stay tuned!
> >> > > > Mikhail
> >> > > >
> >> > >
> >> >
> >> >
> >> >
> >> > --
> >> > Thanks,
> >> > Michael Antonov
> >> >
> >>
> >
> >
> >
> > --
> > Thanks,
> > Michael Antonov
> >
>
>
>
> --
> Thanks,
> Michael Antonov
>
Re: State of 1.3 first RC
Posted by Mikhail Antonov <ol...@gmail.com>.
All rights guys, thanks for the pings and patience! Status update:
- all patches listed in the first email on the thread have been committed
- in addition to those following fixes will go in (good catches!)
* HBASE-16931 Setting cell's seqId to zero in compaction flow might
cause RS down (committed)
* HBASE-16960 RegionServer hang when aborting (committed to master and
branch-1, pending commit to branch-1.3).
With that I'm planning to prepare the RC. The other patches pinged upon
would be in 1.3.1 or 1.4.0.
Speak up if I missed some pending fixes you consider critical.
Thanks!
-Mikhail
On Tue, Nov 1, 2016 at 10:46 AM, Mikhail Antonov <ol...@gmail.com>
wrote:
> Looking - thanks!
>
> -M
>
> On Tue, Nov 1, 2016 at 10:03 AM, Yu Li <ca...@gmail.com> wrote:
>
>> Another critical one for your attention sir (patch ready for commit):
>> https://issues.apache.org/jira/browse/HBASE-16960
>>
>> Best Regards,
>> Yu
>>
>> On 31 October 2016 at 16:59, Mikhail Antonov <ol...@gmail.com>
>> wrote:
>>
>> > Hi Yu,
>> >
>> > sorry I missed that and thanks for reaching out. Let me have a look now
>> and
>> > comment on the jira.
>> >
>> > -Mikhail
>> >
>> > On Mon, Oct 31, 2016 at 12:30 AM, Yu Li <ca...@gmail.com> wrote:
>> >
>> > > I think HBASE-16931 is also a critical issue (oops also in compaction
>> > > area...) and should go in 1.3.0, pinged several times there but
>> failed to
>> > > draw your attention boss (smile). Mind take a look? Thanks.
>> > >
>> > > Btw, is there any ETA for 1.3.0 yet?
>> > >
>> > > Best Regards,
>> > > Yu
>> > >
>> > > On 29 October 2016 at 08:17, Mikhail Antonov <ol...@gmail.com>
>> > wrote:
>> > >
>> > > > Guys,
>> > > >
>> > > > I've been working on getting the first RC out this week and want to
>> > share
>> > > > an update on the current state of the things.
>> > > >
>> > > > - HBASE-16951 - was found and fixed, when we don't include
>> > > > hbase-archetypes in the tarball assembly
>> > > > - HBASE-16570 - optimization that caused a regression in the
>> balancer,
>> > > it
>> > > > was reverted from 1.3. Once fixed that would go to 1.4 and perhaps
>> > 1.3.1.
>> > > > - HBASE-16743 - flaky TestSimpleRpcScheduler, fixed now (I ran
>> several
>> > > > cycles of 50 iterations each and all have passed).
>> > > >
>> > > > Now, there's presently one remaining blocker - HBASE-16964, when we
>> > don't
>> > > > properly handle archiving of store files after compactions (thanks
>> Gary
>> > > for
>> > > > digging into that!).
>> > > > See the jira for details.
>> > > >
>> > > > Also, as a note - the above is a third major bug around compactions
>> > that
>> > > we
>> > > > saw on 1.3 during our testing last month (see also HBASE-16788
>> > > > and HBASE-16754)..
>> > > > So during the upcoming RC testing let's try to stress this area.
>> > > >
>> > > > Until 1.3 is out let's consider branch-1.3 to be in a lockdown for
>> new
>> > > > changes. Ping me if you think there is something that I'm missing
>> that
>> > > > should go out in 1.3. Otherwise,
>> > > > important but non-blocker fixes in flight would go in 1.3.1.
>> > > >
>> > > > Stay tuned!
>> > > > Mikhail
>> > > >
>> > >
>> >
>> >
>> >
>> > --
>> > Thanks,
>> > Michael Antonov
>> >
>>
>
>
>
> --
> Thanks,
> Michael Antonov
>
--
Thanks,
Michael Antonov
Re: State of 1.3 first RC
Posted by Mikhail Antonov <ol...@gmail.com>.
Looking - thanks!
-M
On Tue, Nov 1, 2016 at 10:03 AM, Yu Li <ca...@gmail.com> wrote:
> Another critical one for your attention sir (patch ready for commit):
> https://issues.apache.org/jira/browse/HBASE-16960
>
> Best Regards,
> Yu
>
> On 31 October 2016 at 16:59, Mikhail Antonov <ol...@gmail.com> wrote:
>
> > Hi Yu,
> >
> > sorry I missed that and thanks for reaching out. Let me have a look now
> and
> > comment on the jira.
> >
> > -Mikhail
> >
> > On Mon, Oct 31, 2016 at 12:30 AM, Yu Li <ca...@gmail.com> wrote:
> >
> > > I think HBASE-16931 is also a critical issue (oops also in compaction
> > > area...) and should go in 1.3.0, pinged several times there but failed
> to
> > > draw your attention boss (smile). Mind take a look? Thanks.
> > >
> > > Btw, is there any ETA for 1.3.0 yet?
> > >
> > > Best Regards,
> > > Yu
> > >
> > > On 29 October 2016 at 08:17, Mikhail Antonov <ol...@gmail.com>
> > wrote:
> > >
> > > > Guys,
> > > >
> > > > I've been working on getting the first RC out this week and want to
> > share
> > > > an update on the current state of the things.
> > > >
> > > > - HBASE-16951 - was found and fixed, when we don't include
> > > > hbase-archetypes in the tarball assembly
> > > > - HBASE-16570 - optimization that caused a regression in the
> balancer,
> > > it
> > > > was reverted from 1.3. Once fixed that would go to 1.4 and perhaps
> > 1.3.1.
> > > > - HBASE-16743 - flaky TestSimpleRpcScheduler, fixed now (I ran
> several
> > > > cycles of 50 iterations each and all have passed).
> > > >
> > > > Now, there's presently one remaining blocker - HBASE-16964, when we
> > don't
> > > > properly handle archiving of store files after compactions (thanks
> Gary
> > > for
> > > > digging into that!).
> > > > See the jira for details.
> > > >
> > > > Also, as a note - the above is a third major bug around compactions
> > that
> > > we
> > > > saw on 1.3 during our testing last month (see also HBASE-16788
> > > > and HBASE-16754)..
> > > > So during the upcoming RC testing let's try to stress this area.
> > > >
> > > > Until 1.3 is out let's consider branch-1.3 to be in a lockdown for
> new
> > > > changes. Ping me if you think there is something that I'm missing
> that
> > > > should go out in 1.3. Otherwise,
> > > > important but non-blocker fixes in flight would go in 1.3.1.
> > > >
> > > > Stay tuned!
> > > > Mikhail
> > > >
> > >
> >
> >
> >
> > --
> > Thanks,
> > Michael Antonov
> >
>
--
Thanks,
Michael Antonov
Re: State of 1.3 first RC
Posted by Yu Li <ca...@gmail.com>.
Another critical one for your attention sir (patch ready for commit):
https://issues.apache.org/jira/browse/HBASE-16960
Best Regards,
Yu
On 31 October 2016 at 16:59, Mikhail Antonov <ol...@gmail.com> wrote:
> Hi Yu,
>
> sorry I missed that and thanks for reaching out. Let me have a look now and
> comment on the jira.
>
> -Mikhail
>
> On Mon, Oct 31, 2016 at 12:30 AM, Yu Li <ca...@gmail.com> wrote:
>
> > I think HBASE-16931 is also a critical issue (oops also in compaction
> > area...) and should go in 1.3.0, pinged several times there but failed to
> > draw your attention boss (smile). Mind take a look? Thanks.
> >
> > Btw, is there any ETA for 1.3.0 yet?
> >
> > Best Regards,
> > Yu
> >
> > On 29 October 2016 at 08:17, Mikhail Antonov <ol...@gmail.com>
> wrote:
> >
> > > Guys,
> > >
> > > I've been working on getting the first RC out this week and want to
> share
> > > an update on the current state of the things.
> > >
> > > - HBASE-16951 - was found and fixed, when we don't include
> > > hbase-archetypes in the tarball assembly
> > > - HBASE-16570 - optimization that caused a regression in the balancer,
> > it
> > > was reverted from 1.3. Once fixed that would go to 1.4 and perhaps
> 1.3.1.
> > > - HBASE-16743 - flaky TestSimpleRpcScheduler, fixed now (I ran several
> > > cycles of 50 iterations each and all have passed).
> > >
> > > Now, there's presently one remaining blocker - HBASE-16964, when we
> don't
> > > properly handle archiving of store files after compactions (thanks Gary
> > for
> > > digging into that!).
> > > See the jira for details.
> > >
> > > Also, as a note - the above is a third major bug around compactions
> that
> > we
> > > saw on 1.3 during our testing last month (see also HBASE-16788
> > > and HBASE-16754)..
> > > So during the upcoming RC testing let's try to stress this area.
> > >
> > > Until 1.3 is out let's consider branch-1.3 to be in a lockdown for new
> > > changes. Ping me if you think there is something that I'm missing that
> > > should go out in 1.3. Otherwise,
> > > important but non-blocker fixes in flight would go in 1.3.1.
> > >
> > > Stay tuned!
> > > Mikhail
> > >
> >
>
>
>
> --
> Thanks,
> Michael Antonov
>
Re: State of 1.3 first RC
Posted by Mikhail Antonov <ol...@gmail.com>.
Hi Yu,
sorry I missed that and thanks for reaching out. Let me have a look now and
comment on the jira.
-Mikhail
On Mon, Oct 31, 2016 at 12:30 AM, Yu Li <ca...@gmail.com> wrote:
> I think HBASE-16931 is also a critical issue (oops also in compaction
> area...) and should go in 1.3.0, pinged several times there but failed to
> draw your attention boss (smile). Mind take a look? Thanks.
>
> Btw, is there any ETA for 1.3.0 yet?
>
> Best Regards,
> Yu
>
> On 29 October 2016 at 08:17, Mikhail Antonov <ol...@gmail.com> wrote:
>
> > Guys,
> >
> > I've been working on getting the first RC out this week and want to share
> > an update on the current state of the things.
> >
> > - HBASE-16951 - was found and fixed, when we don't include
> > hbase-archetypes in the tarball assembly
> > - HBASE-16570 - optimization that caused a regression in the balancer,
> it
> > was reverted from 1.3. Once fixed that would go to 1.4 and perhaps 1.3.1.
> > - HBASE-16743 - flaky TestSimpleRpcScheduler, fixed now (I ran several
> > cycles of 50 iterations each and all have passed).
> >
> > Now, there's presently one remaining blocker - HBASE-16964, when we don't
> > properly handle archiving of store files after compactions (thanks Gary
> for
> > digging into that!).
> > See the jira for details.
> >
> > Also, as a note - the above is a third major bug around compactions that
> we
> > saw on 1.3 during our testing last month (see also HBASE-16788
> > and HBASE-16754)..
> > So during the upcoming RC testing let's try to stress this area.
> >
> > Until 1.3 is out let's consider branch-1.3 to be in a lockdown for new
> > changes. Ping me if you think there is something that I'm missing that
> > should go out in 1.3. Otherwise,
> > important but non-blocker fixes in flight would go in 1.3.1.
> >
> > Stay tuned!
> > Mikhail
> >
>
--
Thanks,
Michael Antonov
Re: State of 1.3 first RC
Posted by Yu Li <ca...@gmail.com>.
I think HBASE-16931 is also a critical issue (oops also in compaction
area...) and should go in 1.3.0, pinged several times there but failed to
draw your attention boss (smile). Mind take a look? Thanks.
Btw, is there any ETA for 1.3.0 yet?
Best Regards,
Yu
On 29 October 2016 at 08:17, Mikhail Antonov <ol...@gmail.com> wrote:
> Guys,
>
> I've been working on getting the first RC out this week and want to share
> an update on the current state of the things.
>
> - HBASE-16951 - was found and fixed, when we don't include
> hbase-archetypes in the tarball assembly
> - HBASE-16570 - optimization that caused a regression in the balancer, it
> was reverted from 1.3. Once fixed that would go to 1.4 and perhaps 1.3.1.
> - HBASE-16743 - flaky TestSimpleRpcScheduler, fixed now (I ran several
> cycles of 50 iterations each and all have passed).
>
> Now, there's presently one remaining blocker - HBASE-16964, when we don't
> properly handle archiving of store files after compactions (thanks Gary for
> digging into that!).
> See the jira for details.
>
> Also, as a note - the above is a third major bug around compactions that we
> saw on 1.3 during our testing last month (see also HBASE-16788
> and HBASE-16754)..
> So during the upcoming RC testing let's try to stress this area.
>
> Until 1.3 is out let's consider branch-1.3 to be in a lockdown for new
> changes. Ping me if you think there is something that I'm missing that
> should go out in 1.3. Otherwise,
> important but non-blocker fixes in flight would go in 1.3.1.
>
> Stay tuned!
> Mikhail
>