https://docs.google.com/document/d/1FkP2dDE4zzH1ou89_-qpW6-7dtBj9XdMRGjFnnLGrTI
I'm also going to try to attach it as a txt file, but I'm not sure if
attachments will be scrubbed.
--
Christopher L Tubbs II
http://gravatar.com/ctubbsii
On Thu, Jun 6, 2013 at 2:12 PM, Jonathan Hsieh <[email protected]> wrote:
> Hi,
>
> Can someone share the document produced from yesterday to the mailing list?
> (the link to the hangout seems to just create a new session!)
>
> Thanks,
> Jon.
>
>
> On Wed, Jun 5, 2013 at 11:33 AM, Billie Rinaldi
> <[email protected]>wrote:
>
>> irc.freenode.net
>>
>>
>> On Wed, Jun 5, 2013 at 11:30 AM, David Medinets <[email protected]
>> >wrote:
>>
>> > I'm heading home to hangout now ... Is there some irc server to use for
>> > #accumulo?
>> >
>> >
>> > On Wed, Jun 5, 2013 at 12:25 PM, Billie Rinaldi <
>> [email protected]
>> > >wrote:
>> >
>> > > In case there are technical difficulties with the hangout, I recommend
>> > that
>> > > people also join #accumulo.
>> > >
>> > >
>> > > On Wed, Jun 5, 2013 at 9:04 AM, Christopher <[email protected]>
>> wrote:
>> > >
>> > > > At 1430EDT, right?
>> > > >
>> > > > --
>> > > > Christopher L Tubbs II
>> > > > http://gravatar.com/ctubbsii
>> > > >
>> > > >
>> > > > On Wed, Jun 5, 2013 at 11:05 AM, Billie Rinaldi
>> > > > <[email protected]> wrote:
>> > > > > Let's try this:
>> > > > >
>> > > >
>> > >
>> >
>> https://plus.google.com/hangouts/_/c96631ed67e4add15f0f84ab8ae65ac4aaef6204
>> > > > >
>> > > > > Billie
>> > > > >
>> > > > >
>> > > > > On Wed, Jun 5, 2013 at 7:54 AM, Jonathan Hsieh <[email protected]>
>> > > wrote:
>> > > > >
>> > > > >> How would I attend this -- is here a link I can go to?
>> > > > >>
>> > > > >> Thanks!
>> > > > >> Jon.
>> > > > >>
>> > > > >>
>> > > > >> On Wed, May 29, 2013 at 12:27 PM, Christopher <
>> [email protected]>
>> > > > wrote:
>> > > > >>
>> > > > >> > June 5 makes more sense ;)
>> > > > >> >
>> > > > >> > +1
>> > > > >> >
>> > > > >> > --
>> > > > >> > Christopher L Tubbs II
>> > > > >> > http://gravatar.com/ctubbsii
>> > > > >> >
>> > > > >> >
>> > > > >> > On Wed, May 29, 2013 at 3:26 PM, Corey Nolet <[email protected]
>> >
>> > > > wrote:
>> > > > >> > > Wednesday, June 5. 1430 EDT, correct?
>> > > > >> > >
>> > > > >> > > +1
>> > > > >> > >
>> > > > >> > >
>> > > > >> > > On Wed, May 29, 2013 at 3:23 PM, Christopher <
>> > [email protected]
>> > > >
>> > > > >> > wrote:
>> > > > >> > >
>> > > > >> > >> It is past 1430 EDT on Wednesday. Is this ongoing?
>> > > > >> > >>
>> > > > >> > >> --
>> > > > >> > >> Christopher L Tubbs II
>> > > > >> > >> http://gravatar.com/ctubbsii
>> > > > >> > >>
>> > > > >> > >>
>> > > > >> > >> On Wed, May 29, 2013 at 2:03 PM, Adam Fuchs <
>> [email protected]
>> > >
>> > > > >> wrote:
>> > > > >> > >> > I second 1430 EDT on Wednesday.
>> > > > >> > >> >
>> > > > >> > >> > Adam
>> > > > >> > >> >
>> > > > >> > >> >
>> > > > >> > >> > On Wed, May 29, 2013 at 1:53 PM, Billie Rinaldi <
>> > > > >> > >> [email protected]>wrote:
>> > > > >> > >> >
>> > > > >> > >> >> I can't do it then. I counter-propose Wednesday at 1430
>> ET.
>> > > > >> Two-ish
>> > > > >> > >> hours
>> > > > >> > >> >> sounds reasonable.
>> > > > >> > >> >>
>> > > > >> > >> >> Billie
>> > > > >> > >> >>
>> > > > >> > >> >>
>> > > > >> > >> >> On Wed, May 29, 2013 at 10:46 AM, Adam Fuchs <
>> > > [email protected]
>> > > > >
>> > > > >> > wrote:
>> > > > >> > >> >>
>> > > > >> > >> >> > A hangout is best for me too, but we will have to make
>> > sure
>> > > we
>> > > > >> have
>> > > > >> > >> >> enough
>> > > > >> > >> >> > capacity for everyone who is interested. Let's pick a
>> date
>> > > > first
>> > > > >> > and
>> > > > >> > >> see
>> > > > >> > >> >> > how many people will be joining us. I propose Wednesday
>> > > > >> afternoon,
>> > > > >> > >> >> starting
>> > > > >> > >> >> > at 1300 EDT and going approximately two hours (or until
>> > > we're
>> > > > >> > done).
>> > > > >> > >> >> >
>> > > > >> > >> >> > Adam
>> > > > >> > >> >> >
>> > > > >> > >> >> >
>> > > > >> > >> >> >
>> > > > >> > >> >> > On Tue, May 28, 2013 at 6:57 PM, Corey Nolet <
>> > > > [email protected]>
>> > > > >> > >> wrote:
>> > > > >> > >> >> >
>> > > > >> > >> >> > > I vote google hangout also.
>> > > > >> > >> >> > > On May 28, 2013 4:21 PM, "Christopher" <
>> > > [email protected]
>> > > > >
>> > > > >> > wrote:
>> > > > >> > >> >> > >
>> > > > >> > >> >> > > > Google Hangout works for me.
>> > > > >> > >> >> > > >
>> > > > >> > >> >> > > > --
>> > > > >> > >> >> > > > Christopher L Tubbs II
>> > > > >> > >> >> > > > http://gravatar.com/ctubbsii
>> > > > >> > >> >> > > >
>> > > > >> > >> >> > > >
>> > > > >> > >> >> > > > On Tue, May 28, 2013 at 2:51 PM, John Vines <
>> > > > >> [email protected]>
>> > > > >> > >> >> wrote:
>> > > > >> > >> >> > > > > I'm a fan of using google hangouts because then
>> the
>> > > > session
>> > > > >> > >> could
>> > > > >> > >> >> be
>> > > > >> > >> >> > > > > recorded and thrown up on youtube as reference.
>> > > > >> > >> >> > > > >
>> > > > >> > >> >> > > > >
>> > > > >> > >> >> > > > > On Tue, May 28, 2013 at 2:44 PM, David Medinets <
>> > > > >> > >> >> > > > [email protected]>wrote:
>> > > > >> > >> >> > > > >
>> > > > >> > >> >> > > > >> Where? Is there a central location to people?
>> > > > >> > >> >> > > > >>
>> > > > >> > >> >> > > > >>
>> > > > >> > >> >> > > > >> On Tue, May 28, 2013 at 1:28 PM, Adam Fuchs <
>> > > > >> > [email protected]
>> > > > >> > >> >
>> > > > >> > >> >> > > wrote:
>> > > > >> > >> >> > > > >>
>> > > > >> > >> >> > > > >> > I'm interested. Tuesday's no good for me, but
>> my
>> > > > other
>> > > > >> > days
>> > > > >> > >> are
>> > > > >> > >> >> > > fairly
>> > > > >> > >> >> > > > >> > flexible.
>> > > > >> > >> >> > > > >> >
>> > > > >> > >> >> > > > >> > Adam
>> > > > >> > >> >> > > > >> >
>> > > > >> > >> >> > > > >> >
>> > > > >> > >> >> > > > >> >
>> > > > >> > >> >> > > > >> > On Tue, May 28, 2013 at 12:37 PM, Billie
>> Rinaldi
>> > > > >> > >> >> > > > >> > <[email protected]>wrote:
>> > > > >> > >> >> > > > >> >
>> > > > >> > >> >> > > > >> > > Would anyone be interested in getting
>> together
>> > > next
>> > > > >> > week in
>> > > > >> > >> >> > person
>> > > > >> > >> >> > > > or
>> > > > >> > >> >> > > > >> via
>> > > > >> > >> >> > > > >> > > hangout to discuss 1.6 features?
>> > > > >> > >> >> > > > >> > >
>> > > > >> > >> >> > > > >> > > If we have such a discussion, decisions will
>> > not
>> > > be
>> > > > >> made
>> > > > >> > >> and
>> > > > >> > >> >> > notes
>> > > > >> > >> >> > > > will
>> > > > >> > >> >> > > > >> > be
>> > > > >> > >> >> > > > >> > > posted on the mailing list.
>> > > > >> > >> >> > > > >> > >
>> > > > >> > >> >> > > > >> > > Billie
>> > > > >> > >> >> > > > >> > >
>> > > > >> > >> >> > > > >> >
>> > > > >> > >> >> > > > >>
>> > > > >> > >> >> > > >
>> > > > >> > >> >> > >
>> > > > >> > >> >> >
>> > > > >> > >> >>
>> > > > >> > >>
>> > > > >> >
>> > > > >>
>> > > > >>
>> > > > >>
>> > > > >> --
>> > > > >> // Jonathan Hsieh (shay)
>> > > > >> // Software Engineer, Cloudera
>> > > > >> // [email protected]
>> > > > >>
>> > > >
>> > >
>> >
>>
>
>
>
> --
> // Jonathan Hsieh (shay)
> // Software Engineer, Cloudera
> // [email protected]
Agenda:
1. Jira Use
a. Developer guide on ticket ownership
b. Developer guide link to newbie ticket
c. Unassign tickets not being worked on
2. Release Cycles
a. history
Early on, releases were every 6 months. 1.5 was more like 14
months. We would like to move back to more like a 6 month major release
schedule.
Too many features came in close to the feature freeze date.
Everyone was guilty!
b. future
might need to have a “release master” that makes the hard
decisions
backporting features can be troublesome
pick a set of milestones and a rough schedule early in the
process. milestones should include feature freeze, design review, code freeze,
and rc voting?
define more clearly that code freeze is when coding becomes bug
fixes only
define feature freeze to include documentation, completeness,
etc.
3. GIT
a. benefits
b. procedures / branches
not much to discuss here - mailing list thread is moving
towards consensus
4. 1.6 Features
\000 Documentation (ACCUMULO-1490)
What: better docs, consolidation (into what?)
Who: Sean coordinating, everyone doing it
! replication (ticket?)
What: cross-data center replication via copying WAL and
checkpoint strategy. 1. export tables 2. ??? 3. profit?
Who: maybe Keith?
Related to: d -- offlining snapshots
a. Percolator (ACCUMULO-999) (CAS, locality group performance)
What: built on top of Accumulo, but needs support from Accumulo
(compare and set, locality groups). Compactions may need locality group
support, and in-memory map may need locality group support. Needs to support
frequent inserts/deletes into a single locality group. Separate process to do
processing alongside the tablet server.
Who: high priority for Keith
b. Iterator isolation (ACCUMULO-1188) (excessive memory usage, infinite loops,
etc)
What: handle iterators that use all the memory or go into infinite
loops. Detect them first, then defend against them. Look into separate process
for iterators.
Who: high priority for Eric to detect these problems (debugging)
c. RPC abstraction (encryption on line, non-thrift - ACCUMULO-756 ACCUMULO-1009)
What: Isolate use of Thrift, better documentation of comms across
components,ability to use different thrift versions. Possible follow on work to
isolate abstracted layer in separate class loader
Who: Christopher
d. Improved ops (cluster check and recover utility, offline snapshots, rolling
restarts ACCUMULO-361 ACCUMULO-1454)
What: A series of utilities for administering a cluster.
Who: Sean, Christopher, maybe Keith, maybe John
e. Namespaces (to work around namenode issues) ACCUMULO-118
What: Allows Accumulo to use multiple namenodes in HDFS. Needed this
fall by customers. Much of this work is improving compatibility with federated
HDFS (i.e. splaying files across the namenodes while supporting an Accumulo
directory structure)
Who: Primary project for Eric
f. Isolate FATE
What: Make FATE available outside of Accumulo
Who: John
g. Curator switchover (ACCUMULO-715)
What: Leverage curator to take over common Zookeeper
interactions
Who: John?
i. table namespaces (ACCUMULO-802)
What: syntax to express configuration and permissions for
groups of tables instead of one table at a time (iterators, read/write
permissions, etc.). Same API, use * to refer to everything in the namespace.
Who: Christopher and Sean Hickey
j. RFile as the WAL recovery format/alternative recovery method (e.g.
investigate HBASE-7006 distributed log recovery)
What:
Who: not John
k. RFile refactor for {better backwards compatibility, block filtering,
read/write performance improvement}
What:
Who: Adam will flesh these out as a series of tickets --
looking for more people to work on these.
l. In-memory compaction (ACCUMULO-519)
What:
Who: ???
m. Namenode projects: (accumulo as its own namenode ACCUMULO-722)
What: needs more design work, may be unnecessary
Who: no one (not for 1.6)
n. client cleanup / simplification (ACCUMULO-1483)
What: wack Text? Client API artifact.
Who: Christopher and John?
o. power failure testing
What:
http://techblog.netflix.com/2012/07/chaos-monkey-released-into-wild.html
Who: Adam will open ticket
p. platform optimization (support for systems with more hard drives)
What:
Who:
q. accumulo + ambari (ACCUMULO-136, AMBARI-1930)
What: not sure what we’ll need in accumulo for this, maybe just
packaging
Who: Andres and Billie
r. encryption at rest (in motion under c)
What: ACCUMULO-998
Who: Mike Allen
s. RPM/deb attention (ACCUMULO-210)
What:
Who: Christopher and John?
t. writing to multiple walogs, faster walog recovery (ACCUMULO-1083,
1085)
What:
Who: Keith maybe
u. configurator (ACCUMULO-780)
What:
Who: John and Corey
v. monitor / jmx improvements (ACCUMULO-594, ACCUMULO-694)
What:
Who: Supun and Josh
w. make visibility code accessible outside of Accumulo
What: self-explanatory, separate module (w/ACCUMULO-210?)
Who: Christopher
x. accumulo-maven-plugin (ACCUMULO-1030)
What: Create a maven plugin that helps users do integration
testing against Accumulo
Who: Christopher and Corey?
y. separate root table (ACCUMULO-1481)
What: Simplify complex bootstrapping/admin/recovery code
Who: Christopher
z. further polishing ACCUMULO-259 related internals (ACCUMULO-1312)
What: Look for ways to improve and test the additional
pluggable security features
Who:
aa. New shell / command-line utils
Who : Christopher / Mike Drob
ab. Hadoop 3 support (ACCUMULO-1486)?
What: if it’s released before 1.6, we might want to consider
supporting it
Who: Billie and Eric
Actions:
1. Discuss and schedule milestones on the mailing list (design review, feature
freeze, code freeze, release). Propose November timeframe for release.
2. Add more features from the broader community
3. Estimate times to complete each of these features
4. Nominate 1.6 tickets
