HBase Dev Meeeting Agenda

[Jonathan Gray <jg...@streamy.com>]

(mostly from stack)

Pre-meeting Action Items:

+ Review the 0.20.0 list of issues.  There are 50.  Its too many.  We need
to release soon.  Hadoop 0.20.0 is out and we need to have an hbase to run
on it. Please move out all but the critical:
https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&mode=hide&sorter/order=DESC&sorter/field=priority&resolution=-1&pid=12310753&fixfor=12313474

+ Review the latest HBASE-880 API proposal:
https://issues.apache.org/jira/secure/attachment/12406939/HBASE-880_Design_Doc_v5.pdf
HBASE-880 could be punted from hbase 0.20.0 after discussion but there is
strong support for keeping it in.  If we keep it in, this is the proposed
API we're going to have to live with going forward.  Are folks OK w/ that?


Meeting Agenda:

+ Quick status on state of TRUNK

+ Review of the issues filed against 0.20.0:
https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&mode=hide&sorter/order=DESC&sorter/field=priority&resolution=-1&pid=12310753&fixfor=12313474
Lets punt all but the critical.

+ If HBASE-880/1249/1304 survive -- combined it may mean hbase 0.20.0
doesn't ship for at least a month but we'll be addressing long resident
holes in API, more efficient puts and deletes, saving on a rewrite of all
data migrating in a later version, read speed improvements, better memory
allocation patterns, etc... -- then lets figure how to divvy up the work.

+ Should we do a 0.19.2?

+ Who will represent HBase at the Non-relational DB Meetup on June 11, the
day after hadoop summit?  The line-up is:

Cassandra (Avinash Lakshman)
Dynomite (Cliff Moon)
Apache Hbase (TBA)
Hypertable (Doug Judd)
Voldemort (Jay Kreps)


Jon would prefer a call-in, if we can get one set up.  Otherwise we will
Skype with stack as the hub @ michaelstack.  In any case, everyone just
meet in the IRC channel at 2200-UTC so there's no confusion.

Re: HBase Dev Meeeting Agenda

[stack <st...@duboce.net>]

We reviewed issues slated for 0.20.0 moving out a bunch but also bringing in
a few.  Others were recast because of the understanding that comes only when
all interested parties are in the one "room".  Here is the current list:
https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&mode=hide&sorter/order=DESC&sorter/field=priority&resolution=-1&pid=12310753&fixfor=12313474

It was *thought* that a 0.20.0 release candidate could be ready in about two
weeks only for the HBASE-880/1249/1304 suite (new API, new deletes
mechanism, KeyValues over the client/server divide, and a fleet of other
server efficiencies).  The "880-suite", it was thought, would push out the
release at least another 2 to 4 weeks -- meaning hbase 0.20.0 wouldn't
appear till early June.  The implementers, Erik Holstad and JGray, argued
that whether or not we include the 880-suite should be put off till next
friday.  At that time, they'll have a better idea of how long its actually
going to take.  All on the phone were good with that.

We'll make a 0.19.2 release meantime.

No decision as to who represents at the non-relational meetup.  Ryan Rawson
was proposed (Jean-Daniel Cryans and Jon Gray are representing at the Hadoop
Summit, June 10th).

The hadoop PMC approved two new committers to hbase; Ryan Rawson and Nitay
Joffe.  They have yet to have the Apache fairy dust sprinkled on them but
that should happen in the next week or so (welcome aboard lads).

St.Ack
P.S. Sorry the connection was poor for the Skype participants.  Will try to
do better next time.



On Thu, Apr 30, 2009 at 11:35 AM, Jonathan Gray <jg...@streamy.com> wrote:

> (mostly from stack)
>
> Pre-meeting Action Items:
>
> + Review the 0.20.0 list of issues.  There are 50.  Its too many.  We need
> to release soon.  Hadoop 0.20.0 is out and we need to have an hbase to run
> on it. Please move out all but the critical:
>
> https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&mode=hide&sorter/order=DESC&sorter/field=priority&resolution=-1&pid=12310753&fixfor=12313474
>
> + Review the latest HBASE-880 API proposal:
>
> https://issues.apache.org/jira/secure/attachment/12406939/HBASE-880_Design_Doc_v5.pdf
> HBASE-880 could be punted from hbase 0.20.0 after discussion but there is
> strong support for keeping it in.  If we keep it in, this is the proposed
> API we're going to have to live with going forward.  Are folks OK w/ that?
>
>
> Meeting Agenda:
>
> + Quick status on state of TRUNK
>
> + Review of the issues filed against 0.20.0:
>
> https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&mode=hide&sorter/order=DESC&sorter/field=priority&resolution=-1&pid=12310753&fixfor=12313474
> Lets punt all but the critical.
>
> + If HBASE-880/1249/1304 survive -- combined it may mean hbase 0.20.0
> doesn't ship for at least a month but we'll be addressing long resident
> holes in API, more efficient puts and deletes, saving on a rewrite of all
> data migrating in a later version, read speed improvements, better memory
> allocation patterns, etc... -- then lets figure how to divvy up the work.
>
> + Should we do a 0.19.2?
>
> + Who will represent HBase at the Non-relational DB Meetup on June 11, the
> day after hadoop summit?  The line-up is:
>
> Cassandra (Avinash Lakshman)
> Dynomite (Cliff Moon)
> Apache Hbase (TBA)
> Hypertable (Doug Judd)
> Voldemort (Jay Kreps)
>
>
> Jon would prefer a call-in, if we can get one set up.  Otherwise we will
> Skype with stack as the hub @ michaelstack.  In any case, everyone just
> meet in the IRC channel at 2200-UTC so there's no confusion.