replication fix and hbase 1.4.13?

[Whitney Jackson <wh...@gmail.com>]

Hi,

I've been running 1.4.12 with replication and experiencing the "random
region server aborts" as described here:

https://issues.apache.org/jira/browse/HBASE-23169

The underlying problem and fix (woohoo!) seems to be here:

https://issues.apache.org/jira/browse/HBASE-23205

I do see there is a failed jdk7 check noted at the end of 23205. Will that
prevent the fix from getting into 1.4.13? Also, when is 1.4.13 likely to
drop?

On a related note, how does the "stable" label work? I'm running 1.4.12 in
large part because it has that label. But as I discovered it also has this
known bug with a core feature (replication). It seems serious to me but the
urgency on fixing it seems to be low. That surprises me given the
importance of the bug and the "stable" label. Would I be better off keeping
up with the latest and greatest 2.x if my biggest concern is stability?

Whitney

Re: replication fix and hbase 1.4.13?

[Whitney Jackson <wh...@gmail.com>]

Oh I hadn't noticed that change. Thanks for pointing that out. I'll start
looking into upgrading to 2.2.x.

On Mon, Jan 27, 2020 at 10:25 AM Sean Busbey <bu...@apache.org> wrote:

> As a side note, the current "stable" pointer is the 2.2.3 release as
> of like a week ago.
>
> http://www.apache.org/dist/hbase/stable/
>
> it's noted on the download page:
>
> http://hbase.apache.org/downloads.html
>
> it looks like the text at the top of the asf mirroring page hasn't
> been updated and I don't see an announcement to user@. I'll try to got
> both of those oversights cleared up.
>
> On Mon, Jan 27, 2020 at 11:19 AM Whitney Jackson <wh...@gmail.com>
> wrote:
> >
> > Thanks for the reply! Very helpful.
> >
> > > One more thing I'd ask is that you remember that we're all volunteers
> > here.
> >
> > Absolutely. Appreciate the excellent work! I can see how the "stable"
> > question is less cut and dried than it might seem from the outside. And
> > your answer on that is definitely useful in thinking through which
> version
> > to run.
> >
> > Whitney
> >
> >
> > On Mon, Jan 27, 2020 at 9:18 AM Josh Elser <el...@apache.org> wrote:
> >
> > > That fix is already slated for 1.4.13. There would be an exceptional
> > > issue for it to not be included at this point. The "-1" you saw is just
> > > an automated, nightly test which are not always perfect.
> > >
> > > The general cadence for releases is about 1 month. It looks like the
> > > Christmas season got in the way and we're due for 1.4.13 soon. You can
> > > watch the chatter on the dev mailing list to stay in the loop.
> > >
> > > Your question about "stable" is not so easy to answer. It's a label
> > > assigned by the HBase developers to try to give a simple answer to a
> > > very complex question. Developers decide on what we feel is acceptable
> > > to call "stable", and will move it when we think it's
> > > appropriate/necessary. Our goal is to try to make the best experience
> > > for our users to save you all pain/heartache/grief.
> > >
> > > One more thing I'd ask is that you remember that we're all volunteers
> > > here. Day jobs, families, and all sorts of other things can get in the
> > > way. This doesn't mean that something isn't, persay, urgent or critical
> > > to "do" (fix, investigate, release), but just that we don't have enough
> > > bandwidth to do everything we're asked to do.
> > >
> > > Specifically around releases, we're always looking for more people to
> > > help drive the release process. Those who can corral Jira issues, do
> > > testing, and stage release candidates are very welcome and desired to
> > > help make our releases happen on a regular cadence. If you have the
> > > time/resources to help our, let us know on the dev list.
> > >
> > > - Josh
> > >
> > > On 1/27/20 11:04 AM, Whitney Jackson wrote:
> > > > Hi,
> > > >
> > > > I've been running 1.4.12 with replication and experiencing the
> "random
> > > > region server aborts" as described here:
> > > >
> > > > https://issues.apache.org/jira/browse/HBASE-23169
> > > >
> > > > The underlying problem and fix (woohoo!) seems to be here:
> > > >
> > > > https://issues.apache.org/jira/browse/HBASE-23205
> > > >
> > > > I do see there is a failed jdk7 check noted at the end of 23205. Will
> > > that
> > > > prevent the fix from getting into 1.4.13? Also, when is 1.4.13
> likely to
> > > > drop?
> > > >
> > > > On a related note, how does the "stable" label work? I'm running
> 1.4.12
> > > in
> > > > large part because it has that label. But as I discovered it also has
> > > this
> > > > known bug with a core feature (replication). It seems serious to me
> but
> > > the
> > > > urgency on fixing it seems to be low. That surprises me given the
> > > > importance of the bug and the "stable" label. Would I be better off
> > > keeping
> > > > up with the latest and greatest 2.x if my biggest concern is
> stability?
> > > >
> > > > Whitney
> > > >
> > >
>

Re: replication fix and hbase 1.4.13?

[Sean Busbey <bu...@apache.org>]

As a side note, the current "stable" pointer is the 2.2.3 release as
of like a week ago.

http://www.apache.org/dist/hbase/stable/

it's noted on the download page:

http://hbase.apache.org/downloads.html

it looks like the text at the top of the asf mirroring page hasn't
been updated and I don't see an announcement to user@. I'll try to got
both of those oversights cleared up.

On Mon, Jan 27, 2020 at 11:19 AM Whitney Jackson <wh...@gmail.com> wrote:
>
> Thanks for the reply! Very helpful.
>
> > One more thing I'd ask is that you remember that we're all volunteers
> here.
>
> Absolutely. Appreciate the excellent work! I can see how the "stable"
> question is less cut and dried than it might seem from the outside. And
> your answer on that is definitely useful in thinking through which version
> to run.
>
> Whitney
>
>
> On Mon, Jan 27, 2020 at 9:18 AM Josh Elser <el...@apache.org> wrote:
>
> > That fix is already slated for 1.4.13. There would be an exceptional
> > issue for it to not be included at this point. The "-1" you saw is just
> > an automated, nightly test which are not always perfect.
> >
> > The general cadence for releases is about 1 month. It looks like the
> > Christmas season got in the way and we're due for 1.4.13 soon. You can
> > watch the chatter on the dev mailing list to stay in the loop.
> >
> > Your question about "stable" is not so easy to answer. It's a label
> > assigned by the HBase developers to try to give a simple answer to a
> > very complex question. Developers decide on what we feel is acceptable
> > to call "stable", and will move it when we think it's
> > appropriate/necessary. Our goal is to try to make the best experience
> > for our users to save you all pain/heartache/grief.
> >
> > One more thing I'd ask is that you remember that we're all volunteers
> > here. Day jobs, families, and all sorts of other things can get in the
> > way. This doesn't mean that something isn't, persay, urgent or critical
> > to "do" (fix, investigate, release), but just that we don't have enough
> > bandwidth to do everything we're asked to do.
> >
> > Specifically around releases, we're always looking for more people to
> > help drive the release process. Those who can corral Jira issues, do
> > testing, and stage release candidates are very welcome and desired to
> > help make our releases happen on a regular cadence. If you have the
> > time/resources to help our, let us know on the dev list.
> >
> > - Josh
> >
> > On 1/27/20 11:04 AM, Whitney Jackson wrote:
> > > Hi,
> > >
> > > I've been running 1.4.12 with replication and experiencing the "random
> > > region server aborts" as described here:
> > >
> > > https://issues.apache.org/jira/browse/HBASE-23169
> > >
> > > The underlying problem and fix (woohoo!) seems to be here:
> > >
> > > https://issues.apache.org/jira/browse/HBASE-23205
> > >
> > > I do see there is a failed jdk7 check noted at the end of 23205. Will
> > that
> > > prevent the fix from getting into 1.4.13? Also, when is 1.4.13 likely to
> > > drop?
> > >
> > > On a related note, how does the "stable" label work? I'm running 1.4.12
> > in
> > > large part because it has that label. But as I discovered it also has
> > this
> > > known bug with a core feature (replication). It seems serious to me but
> > the
> > > urgency on fixing it seems to be low. That surprises me given the
> > > importance of the bug and the "stable" label. Would I be better off
> > keeping
> > > up with the latest and greatest 2.x if my biggest concern is stability?
> > >
> > > Whitney
> > >
> >

Re: replication fix and hbase 1.4.13?

[Whitney Jackson <wh...@gmail.com>]

Thanks for the reply! Very helpful.

> One more thing I'd ask is that you remember that we're all volunteers
here.

Absolutely. Appreciate the excellent work! I can see how the "stable"
question is less cut and dried than it might seem from the outside. And
your answer on that is definitely useful in thinking through which version
to run.

Whitney


On Mon, Jan 27, 2020 at 9:18 AM Josh Elser <el...@apache.org> wrote:

> That fix is already slated for 1.4.13. There would be an exceptional
> issue for it to not be included at this point. The "-1" you saw is just
> an automated, nightly test which are not always perfect.
>
> The general cadence for releases is about 1 month. It looks like the
> Christmas season got in the way and we're due for 1.4.13 soon. You can
> watch the chatter on the dev mailing list to stay in the loop.
>
> Your question about "stable" is not so easy to answer. It's a label
> assigned by the HBase developers to try to give a simple answer to a
> very complex question. Developers decide on what we feel is acceptable
> to call "stable", and will move it when we think it's
> appropriate/necessary. Our goal is to try to make the best experience
> for our users to save you all pain/heartache/grief.
>
> One more thing I'd ask is that you remember that we're all volunteers
> here. Day jobs, families, and all sorts of other things can get in the
> way. This doesn't mean that something isn't, persay, urgent or critical
> to "do" (fix, investigate, release), but just that we don't have enough
> bandwidth to do everything we're asked to do.
>
> Specifically around releases, we're always looking for more people to
> help drive the release process. Those who can corral Jira issues, do
> testing, and stage release candidates are very welcome and desired to
> help make our releases happen on a regular cadence. If you have the
> time/resources to help our, let us know on the dev list.
>
> - Josh
>
> On 1/27/20 11:04 AM, Whitney Jackson wrote:
> > Hi,
> >
> > I've been running 1.4.12 with replication and experiencing the "random
> > region server aborts" as described here:
> >
> > https://issues.apache.org/jira/browse/HBASE-23169
> >
> > The underlying problem and fix (woohoo!) seems to be here:
> >
> > https://issues.apache.org/jira/browse/HBASE-23205
> >
> > I do see there is a failed jdk7 check noted at the end of 23205. Will
> that
> > prevent the fix from getting into 1.4.13? Also, when is 1.4.13 likely to
> > drop?
> >
> > On a related note, how does the "stable" label work? I'm running 1.4.12
> in
> > large part because it has that label. But as I discovered it also has
> this
> > known bug with a core feature (replication). It seems serious to me but
> the
> > urgency on fixing it seems to be low. That surprises me given the
> > importance of the bug and the "stable" label. Would I be better off
> keeping
> > up with the latest and greatest 2.x if my biggest concern is stability?
> >
> > Whitney
> >
>

Re: replication fix and hbase 1.4.13?

[Josh Elser <el...@apache.org>]

That fix is already slated for 1.4.13. There would be an exceptional 
issue for it to not be included at this point. The "-1" you saw is just 
an automated, nightly test which are not always perfect.

The general cadence for releases is about 1 month. It looks like the 
Christmas season got in the way and we're due for 1.4.13 soon. You can 
watch the chatter on the dev mailing list to stay in the loop.

Your question about "stable" is not so easy to answer. It's a label 
assigned by the HBase developers to try to give a simple answer to a 
very complex question. Developers decide on what we feel is acceptable 
to call "stable", and will move it when we think it's 
appropriate/necessary. Our goal is to try to make the best experience 
for our users to save you all pain/heartache/grief.

One more thing I'd ask is that you remember that we're all volunteers 
here. Day jobs, families, and all sorts of other things can get in the 
way. This doesn't mean that something isn't, persay, urgent or critical 
to "do" (fix, investigate, release), but just that we don't have enough 
bandwidth to do everything we're asked to do.

Specifically around releases, we're always looking for more people to 
help drive the release process. Those who can corral Jira issues, do 
testing, and stage release candidates are very welcome and desired to 
help make our releases happen on a regular cadence. If you have the 
time/resources to help our, let us know on the dev list.

- Josh

On 1/27/20 11:04 AM, Whitney Jackson wrote:
> Hi,
> 
> I've been running 1.4.12 with replication and experiencing the "random
> region server aborts" as described here:
> 
> https://issues.apache.org/jira/browse/HBASE-23169
> 
> The underlying problem and fix (woohoo!) seems to be here:
> 
> https://issues.apache.org/jira/browse/HBASE-23205
> 
> I do see there is a failed jdk7 check noted at the end of 23205. Will that
> prevent the fix from getting into 1.4.13? Also, when is 1.4.13 likely to
> drop?
> 
> On a related note, how does the "stable" label work? I'm running 1.4.12 in
> large part because it has that label. But as I discovered it also has this
> known bug with a core feature (replication). It seems serious to me but the
> urgency on fixing it seems to be low. That surprises me given the
> importance of the bug and the "stable" label. Would I be better off keeping
> up with the latest and greatest 2.x if my biggest concern is stability?
> 
> Whitney
>