Grokbase Groups HBase dev June 2010
FAQ
Hey everyone,

I'd like to start planning towards a development release this month. Here's
my thinking on timeline, working backwards:

*Tues, 6/29 - Hadoop Summit, official first release of dev series*

It would be great to have a release at this point that we think is
moderately usable (ie not significantly less stable than 0.20.n has been).
Perhaps some of the new features (eg replication, coprocessors, etc) would
still be in progress but I think it would be great to be able to talk up a
working installable-today release at the summit.

Thinking backwards, from here:

*Tues, 6/22 - Major patches committed, first "usable" builds w/ new features
*

If all the real development features were committed by 6/22, that would give
us a week to bang on it suss out the bugs. Those of you working on larger
things (eg master fixup), do you think you'll be ready for it to be
committed at this point?

*Tues, 6/15 - Major patches in review*
*
*
Given the scope of some of the changes in the works (particularly thinking
about master work being done at FB right now) I think it would be excellent
if the rest of us could start looking at patches by 6/15. Fine if there are
still known bugs, but this will give us some days to look at the patch and
work in parallel on fixing them.

Do these target dates seem reasonable to everyone?

-Todd

--
Todd Lipcon
Software Engineer, Cloudera

Search Discussions

  • Stack at Jun 4, 2010 at 5:12 am
    Dates look good to me.

    There are too many open issues currently filed against
    http://su.pr/698kib If fellas have a moment, take a look. See what
    can be purged (Are there any we should pull in that are not currently
    in the list?) Adopt those you think you'll be able to work on. Looks
    like a few have wrong priority; e.g. master rewrite is major true, but
    probably should be a blocker.

    Good stuff,
    St.Ack


    On Thu, Jun 3, 2010 at 10:01 PM, Todd Lipcon wrote:
    Hey everyone,

    I'd like to start planning towards a development release this month. Here's
    my thinking on timeline, working backwards:

    *Tues, 6/29 - Hadoop Summit, official first release of dev series*

    It would be great to have a release at this point that we think is
    moderately usable (ie not significantly less stable than 0.20.n has been).
    Perhaps some of the new features (eg replication, coprocessors, etc) would
    still be in progress but I think it would be great to be able to talk up a
    working installable-today release at the summit.

    Thinking backwards, from here:

    *Tues, 6/22 - Major patches committed, first "usable" builds w/ new features
    *

    If all the real development features were committed by 6/22, that would give
    us a week to bang on it suss out the bugs. Those of you working on larger
    things (eg master fixup), do you think you'll be ready for it to be
    committed at this point?

    *Tues, 6/15 - Major patches in review*
    *
    *
    Given the scope of some of the changes in the works (particularly thinking
    about master work being done at FB right now) I think it would be excellent
    if the rest of us could start looking at patches by 6/15. Fine if there are
    still known bugs, but this will give us some days to look at the patch and
    work in parallel on fixing them.

    Do these target dates seem reasonable to everyone?

    -Todd

    --
    Todd Lipcon
    Software Engineer, Cloudera
  • Dhruba Borthakur at Jun 4, 2010 at 7:50 am
    This also means that we should target a Hadoop release from the
    branch-0.20-append, isn't it?

    thanks
    dhruba
    On Thu, Jun 3, 2010 at 10:12 PM, Stack wrote:

    Dates look good to me.

    There are too many open issues currently filed against
    http://su.pr/698kib If fellas have a moment, take a look. See what
    can be purged (Are there any we should pull in that are not currently
    in the list?) Adopt those you think you'll be able to work on. Looks
    like a few have wrong priority; e.g. master rewrite is major true, but
    probably should be a blocker.

    Good stuff,
    St.Ack


    On Thu, Jun 3, 2010 at 10:01 PM, Todd Lipcon wrote:
    Hey everyone,

    I'd like to start planning towards a development release this month. Here's
    my thinking on timeline, working backwards:

    *Tues, 6/29 - Hadoop Summit, official first release of dev series*

    It would be great to have a release at this point that we think is
    moderately usable (ie not significantly less stable than 0.20.n has been).
    Perhaps some of the new features (eg replication, coprocessors, etc) would
    still be in progress but I think it would be great to be able to talk up a
    working installable-today release at the summit.

    Thinking backwards, from here:

    *Tues, 6/22 - Major patches committed, first "usable" builds w/ new features
    *

    If all the real development features were committed by 6/22, that would give
    us a week to bang on it suss out the bugs. Those of you working on larger
    things (eg master fixup), do you think you'll be ready for it to be
    committed at this point?

    *Tues, 6/15 - Major patches in review*
    *
    *
    Given the scope of some of the changes in the works (particularly thinking
    about master work being done at FB right now) I think it would be excellent
    if the rest of us could start looking at patches by 6/15. Fine if there are
    still known bugs, but this will give us some days to look at the patch and
    work in parallel on fixing them.

    Do these target dates seem reasonable to everyone?

    -Todd

    --
    Todd Lipcon
    Software Engineer, Cloudera


    --
    Connect to me at http://www.facebook.com/dhruba
  • Todd Lipcon at Jun 4, 2010 at 7:58 am

    On Fri, Jun 4, 2010 at 12:50 AM, Dhruba Borthakur wrote:

    This also means that we should target a Hadoop release from the
    branch-0.20-append, isn't it?
    Probably, though I'm still a little confused as to who will be testing that
    branch. I plan on concentrating my testing on top of CDH3b2, and I imagine
    you guys are testing based on your internal branch. So we can certainly do
    an append branch release, but I think we'll have to count on community
    testing a bit more.
    On Thu, Jun 3, 2010 at 10:12 PM, Stack wrote:

    Dates look good to me.

    There are too many open issues currently filed against
    http://su.pr/698kib If fellas have a moment, take a look. See what
    can be purged (Are there any we should pull in that are not currently
    in the list?) Adopt those you think you'll be able to work on. Looks
    like a few have wrong priority; e.g. master rewrite is major true, but
    probably should be a blocker.
    I took a swing through some of the blockers to try to triage some down.
    While we have a ton of open issues, we should try to close out the real show
    stoppers first. So long as trunk works generally as well as 0.20.5 I think
    we'll be able to get some users trying out a dev release by offering the
    durability carrot (plus the few big new features we've already got done)

    -Todd
    Good stuff,
    St.Ack


    On Thu, Jun 3, 2010 at 10:01 PM, Todd Lipcon wrote:
    Hey everyone,

    I'd like to start planning towards a development release this month. Here's
    my thinking on timeline, working backwards:

    *Tues, 6/29 - Hadoop Summit, official first release of dev series*

    It would be great to have a release at this point that we think is
    moderately usable (ie not significantly less stable than 0.20.n has been).
    Perhaps some of the new features (eg replication, coprocessors, etc) would
    still be in progress but I think it would be great to be able to talk
    up
    a
    working installable-today release at the summit.

    Thinking backwards, from here:

    *Tues, 6/22 - Major patches committed, first "usable" builds w/ new features
    *

    If all the real development features were committed by 6/22, that would give
    us a week to bang on it suss out the bugs. Those of you working on
    larger
    things (eg master fixup), do you think you'll be ready for it to be
    committed at this point?

    *Tues, 6/15 - Major patches in review*
    *
    *
    Given the scope of some of the changes in the works (particularly thinking
    about master work being done at FB right now) I think it would be excellent
    if the rest of us could start looking at patches by 6/15. Fine if there are
    still known bugs, but this will give us some days to look at the patch and
    work in parallel on fixing them.

    Do these target dates seem reasonable to everyone?

    -Todd

    --
    Todd Lipcon
    Software Engineer, Cloudera


    --
    Connect to me at http://www.facebook.com/dhruba


    --
    Todd Lipcon
    Software Engineer, Cloudera
  • Stack at Jun 4, 2010 at 3:03 pm

    On Fri, Jun 4, 2010 at 12:58 AM, Todd Lipcon wrote:
    On Fri, Jun 4, 2010 at 12:50 AM, Dhruba Borthakur wrote:

    This also means that we should target a Hadoop release from the
    branch-0.20-append, isn't it?
    Probably
    I'd say that sounds great Dhruba. HBase bundles a hadoop. I'd
    imagine the branch-0.20-append is what we'd bundle in this case. Its
    what we'll tell users install, it or optionally cdh3b2.

    St.Ack
  • Jonathan Gray at Jun 5, 2010 at 4:50 pm
    The dates sound good, at least on master side of things.

    I've got a bunch of half-done stuff on the compaction/flush/split side of things that I've been holding off on until the master stuff pushes through. I suspect 6/22 is enough time to finish my stuff, but I may not have all patches up by 6/15. I'll keep that as the target and see how far I can get. I'd really like to at least get the refactor and HBASE-2375 in before the dev release, some of the other changes can wait.

    Also, +1 on doing a dev release of the hadoop branch to ship the dev release with. Part of the way we can get it tested is by putting it out there to the community.

    JG
    -----Original Message-----
    From: Todd Lipcon
    Sent: Thursday, June 03, 2010 10:02 PM
    To: dev
    Subject: Working towards a dev release

    Hey everyone,

    I'd like to start planning towards a development release this month.
    Here's
    my thinking on timeline, working backwards:

    *Tues, 6/29 - Hadoop Summit, official first release of dev series*

    It would be great to have a release at this point that we think is
    moderately usable (ie not significantly less stable than 0.20.n has
    been).
    Perhaps some of the new features (eg replication, coprocessors, etc)
    would
    still be in progress but I think it would be great to be able to talk
    up a
    working installable-today release at the summit.

    Thinking backwards, from here:

    *Tues, 6/22 - Major patches committed, first "usable" builds w/ new
    features
    *

    If all the real development features were committed by 6/22, that would
    give
    us a week to bang on it suss out the bugs. Those of you working on
    larger
    things (eg master fixup), do you think you'll be ready for it to be
    committed at this point?

    *Tues, 6/15 - Major patches in review*
    *
    *
    Given the scope of some of the changes in the works (particularly
    thinking
    about master work being done at FB right now) I think it would be
    excellent
    if the rest of us could start looking at patches by 6/15. Fine if there
    are
    still known bugs, but this will give us some days to look at the patch
    and
    work in parallel on fixing them.

    Do these target dates seem reasonable to everyone?

    -Todd

    --
    Todd Lipcon
    Software Engineer, Cloudera
  • Todd Lipcon at Jun 5, 2010 at 5:41 pm

    On Sat, Jun 5, 2010 at 12:46 PM, Jonathan Gray wrote:

    The dates sound good, at least on master side of things.

    I've got a bunch of half-done stuff on the compaction/flush/split side of
    things that I've been holding off on until the master stuff pushes through.
    I suspect 6/22 is enough time to finish my stuff, but I may not have all
    patches up by 6/15. I'll keep that as the target and see how far I can get.
    I'd really like to at least get the refactor and HBASE-2375 in before the
    dev release, some of the other changes can wait.
    Great. Any chance you could push those works-in-progress off to github? We
    could get some more eyes on it early that way, so review will go faster when
    it's ready to go?

    -Todd

    Also, +1 on doing a dev release of the hadoop branch to ship the dev
    release with. Part of the way we can get it tested is by putting it out
    there to the community.

    JG
    -----Original Message-----
    From: Todd Lipcon
    Sent: Thursday, June 03, 2010 10:02 PM
    To: dev
    Subject: Working towards a dev release

    Hey everyone,

    I'd like to start planning towards a development release this month.
    Here's
    my thinking on timeline, working backwards:

    *Tues, 6/29 - Hadoop Summit, official first release of dev series*

    It would be great to have a release at this point that we think is
    moderately usable (ie not significantly less stable than 0.20.n has
    been).
    Perhaps some of the new features (eg replication, coprocessors, etc)
    would
    still be in progress but I think it would be great to be able to talk
    up a
    working installable-today release at the summit.

    Thinking backwards, from here:

    *Tues, 6/22 - Major patches committed, first "usable" builds w/ new
    features
    *

    If all the real development features were committed by 6/22, that would
    give
    us a week to bang on it suss out the bugs. Those of you working on
    larger
    things (eg master fixup), do you think you'll be ready for it to be
    committed at this point?

    *Tues, 6/15 - Major patches in review*
    *
    *
    Given the scope of some of the changes in the works (particularly
    thinking
    about master work being done at FB right now) I think it would be
    excellent
    if the rest of us could start looking at patches by 6/15. Fine if there
    are
    still known bugs, but this will give us some days to look at the patch
    and
    work in parallel on fixing them.

    Do these target dates seem reasonable to everyone?

    -Todd

    --
    Todd Lipcon
    Software Engineer, Cloudera


    --
    Todd Lipcon
    Software Engineer, Cloudera
  • Jonathan Gray at Jun 6, 2010 at 4:20 pm
    I'll be putting up patches to jira/reviewboard when they're ready for consumption. There's no unified thing that would actually compile and make sense in a github or I would just push it out. All of my compact/split stuff was based on branch and was before the master changes which impact it so have work to do before it's ready for review.

    We are going to break stuff up into smaller chunks so I won't just post a single gigantic patch. That may include some jiras that are just refactors but don't change anything. This will allow actual logic changes to be more concentrated.

    JG
    -----Original Message-----
    From: Todd Lipcon
    Sent: Saturday, June 05, 2010 10:40 AM
    To: dev@hbase.apache.org
    Subject: Re: Working towards a dev release
    On Sat, Jun 5, 2010 at 12:46 PM, Jonathan Gray wrote:

    The dates sound good, at least on master side of things.

    I've got a bunch of half-done stuff on the compaction/flush/split side of
    things that I've been holding off on until the master stuff pushes through.
    I suspect 6/22 is enough time to finish my stuff, but I may not have all
    patches up by 6/15. I'll keep that as the target and see how far I can get.
    I'd really like to at least get the refactor and HBASE-2375 in
    before the
    dev release, some of the other changes can wait.
    Great. Any chance you could push those works-in-progress off to github?
    We
    could get some more eyes on it early that way, so review will go faster
    when
    it's ready to go?

    -Todd

    Also, +1 on doing a dev release of the hadoop branch to ship the dev
    release with. Part of the way we can get it tested is by putting it out
    there to the community.

    JG
    -----Original Message-----
    From: Todd Lipcon
    Sent: Thursday, June 03, 2010 10:02 PM
    To: dev
    Subject: Working towards a dev release

    Hey everyone,

    I'd like to start planning towards a development release this
    month.
    Here's
    my thinking on timeline, working backwards:

    *Tues, 6/29 - Hadoop Summit, official first release of dev series*

    It would be great to have a release at this point that we think is
    moderately usable (ie not significantly less stable than 0.20.n has
    been).
    Perhaps some of the new features (eg replication, coprocessors,
    etc)
    would
    still be in progress but I think it would be great to be able to
    talk
    up a
    working installable-today release at the summit.

    Thinking backwards, from here:

    *Tues, 6/22 - Major patches committed, first "usable" builds w/ new
    features
    *

    If all the real development features were committed by 6/22, that
    would
    give
    us a week to bang on it suss out the bugs. Those of you working on
    larger
    things (eg master fixup), do you think you'll be ready for it to be
    committed at this point?

    *Tues, 6/15 - Major patches in review*
    *
    *
    Given the scope of some of the changes in the works (particularly
    thinking
    about master work being done at FB right now) I think it would be
    excellent
    if the rest of us could start looking at patches by 6/15. Fine if
    there
    are
    still known bugs, but this will give us some days to look at the
    patch
    and
    work in parallel on fixing them.

    Do these target dates seem reasonable to everyone?

    -Todd

    --
    Todd Lipcon
    Software Engineer, Cloudera


    --
    Todd Lipcon
    Software Engineer, Cloudera

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupdev @
categorieshbase, hadoop
postedJun 4, '10 at 5:02a
activeJun 6, '10 at 4:20p
posts8
users4
websitehbase.apache.org

People

Translate

site design / logo © 2021 Grokbase