lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tomás Fernández Löbbe <tomasflo...@gmail.com>
Subject Re: Lucene/Solr 7.5
Date Wed, 12 Sep 2018 22:10:28 GMT
Hi Jim,
I'd like to commit SOLR-12766 to 7.5. SOLR-11881 added retries for internal
requests, but the backoff time in cases with multiple updates can become
big, and cause clients to timeout. The change is minimal, just backoff once
for a retry batch instead of for every doc.

I'm testing a patch and plan to commit later today, if there aren't any
issues or objections.

On Wed, Sep 12, 2018 at 5:39 AM jim ferenczi <jim.ferenczi@gmail.com> wrote:

> Thanks !
>
> Le mer. 12 sept. 2018 à 11:49, Adrien Grand <jpountz@gmail.com> a écrit :
>
>> Hey Jim,
>>
>> I added you to the hudson-jobadmin group so that you can do it next time.
>>
>> Steve, thanks for taking care of setting up the builds!
>>
>> Le mar. 11 sept. 2018 à 17:32, jim ferenczi <jim.ferenczi@gmail.com> a
>> écrit :
>>
>>> No worries at all Cassandra. What do you think of building the first RC
>>> on Friday and start the vote on Monday next week ? This will leave some
>>> room to finish the missing bits.
>>> Could someone help to setup the Jenkins releases build ? It seems that I
>>> cannot create jobs with my account.
>>>
>>> Le mar. 11 sept. 2018 à 14:08, Cassandra Targett <casstargett@gmail.com>
>>> a écrit :
>>>
>>>> Sorry, Jim, I should have replied yesterday about the state of things
>>>> with the Ref Guide - it's close. I'm doing the last bit of big review I
>>>> need to do and am nearly done with that, then I have a couple more small
>>>> things done (including SOLR-12763 which I just created since I forgot to
do
>>>> it earlier). My goal is to be done by the end of my day today so you could
>>>> do the RC tomorrow, but who knows what the day will bring work-wise, so
>>>> I'll send another mail at the end of the day my time to let you know for
>>>> sure.
>>>>
>>>> On Mon, Sep 10, 2018 at 9:07 AM jim ferenczi <jim.ferenczi@gmail.com>
>>>> wrote:
>>>>
>>>>> I just fixed the invalid version (7.5.1) that I added in master and
>>>>> 7x. The next version on these branches should be 7.6.0, sorry for the
noise.
>>>>>
>>>>> Le lun. 10 sept. 2018 à 09:26, jim ferenczi <jim.ferenczi@gmail.com>
>>>>> a écrit :
>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> Feature freeze for 7.5 has started, I just created a branch_7_5.:
>>>>>>
>>>>>> * No new features may be committed to the branch.
>>>>>> * Documentation patches, build patches and serious bug fixes may
be
>>>>>> committed to the branch. However, you should submit all patches you
want to
>>>>>> commit to Jira first to give others the chance to review and possibly
vote
>>>>>> against the patch. Keep in mind that it is our main intention to
keep the
>>>>>> branch as stable as possible.
>>>>>> * All patches that are intended for the branch should first be
>>>>>> committed to the unstable branch, merged into the stable branch,
and then
>>>>>> into the current release branch.
>>>>>> * Normal unstable and stable branch development may continue as
>>>>>> usual. However, if you plan to commit a big change to the unstable
branch
>>>>>> while the branch feature freeze is in effect, think twice: can't
the
>>>>>> addition wait a couple more days? Merges of bug fixes into the branch
may
>>>>>> become more difficult.
>>>>>> * Only Jira issues with Fix version "7.5" and priority "Blocker"
will
>>>>>> delay a release candidate build.
>>>>>>
>>>>>> I'll create the first RC later this week depending on the status
of
>>>>>> the Solr ref guide. Cassandra, can you update the status when you
think
>>>>>> that the ref guide is ready (no rush just a reminder that we need
to sync
>>>>>> during this release ;) ) ?
>>>>>>
>>>>>> Cheers,
>>>>>> Jim
>>>>>>
>>>>>> Le mer. 5 sept. 2018 à 17:57, Erick Erickson <erickerickson@gmail.com>
>>>>>> a écrit :
>>>>>>
>>>>>>> Great, thanks!
>>>>>>> On Wed, Sep 5, 2018 at 8:44 AM jim ferenczi <jim.ferenczi@gmail.com>
>>>>>>> wrote:
>>>>>>> >
>>>>>>> > Sure it can wait a few days. Let's cut the branch next Monday
and
>>>>>>> we can sync with Cassandra to create the first RC when the ref
guide is
>>>>>>> ready.
>>>>>>> >
>>>>>>> > Le mer. 5 sept. 2018 à 17:27, Erick Erickson <
>>>>>>> erickerickson@gmail.com> a écrit :
>>>>>>> >>
>>>>>>> >> Jim:
>>>>>>> >>
>>>>>>> >> I know it's the 11th hour, but WDYT about cutting the
branch next
>>>>>>> >> Monday? We see a flurry of activity (announcing a release
does
>>>>>>> >> that....) and waiting to cut the branch might be easiest
all
>>>>>>> 'round.
>>>>>>> >>
>>>>>>> >> Up to you of course, I can backport the test fixes I'd
like for
>>>>>>> >> instance and I'd like to get the upgraded ZooKeeper
in 7.5.
>>>>>>> >>
>>>>>>> >> Erick
>>>>>>> >> On Tue, Sep 4, 2018 at 1:04 PM Cassandra Targett <
>>>>>>> casstargett@gmail.com> wrote:
>>>>>>> >> >
>>>>>>> >> > It's not so much the building of the RC as giving
the content a
>>>>>>> detailed editorial review.
>>>>>>> >> >
>>>>>>> >> > The build/release process itself is well-documented
and
>>>>>>> published with every Ref Guide:
>>>>>>> https://lucene.apache.org/solr/guide/how-to-contribute.html#building-publishing-the-guide.
>>>>>>> It was designed from the artifact process, so it's nearly identical
as a
>>>>>>> process. It's really barely a burden.
>>>>>>> >> >
>>>>>>> >> > In terms of preparing the content, there are a
number of things
>>>>>>> I do:
>>>>>>> >> >
>>>>>>> >> > First, I try to ensure that every issue in CHANGES.txt
that
>>>>>>> should be documented has been documented. That involves an intensive
review
>>>>>>> of CHANGES.txt and a comparison with commits to find what might
be missing,
>>>>>>> then chasing people down to see if they intend to make changes
or not.
>>>>>>> Assuming the person responds, then it's waiting for them to get
their stuff
>>>>>>> done. This is usually about 2-3 days of effort, before the waiting
around
>>>>>>> for answers and/or commits.
>>>>>>> >> >
>>>>>>> >> > Then I review every commit and read it for clarity
and correct
>>>>>>> English usage. Does it fit where someone put it? Does it explain
what the
>>>>>>> author is hoping it explains? Also, many of our authors are not
native
>>>>>>> English writers, and deserve the assistance of an editor to help
put their
>>>>>>> work in the best possible light. In some cases, I feel I should
extensively
>>>>>>> edit the contribution, which occasionally involves also immersing
myself
>>>>>>> into the change itself. This is another 2-4 days of effort.
>>>>>>> >> >
>>>>>>> >> > Then there's this list of problems people commit
all the time,
>>>>>>> many of which I can often resolve reasonably quickly with find/replace:
>>>>>>> >> >
>>>>>>> >> > - sentences that don't end in periods
>>>>>>> >> > - inconsistency with instances of "i.e.," and "e.g.,"
(not
>>>>>>> "i.e.", "ie:", "IE", etc.)
>>>>>>> >> > - no spaces between words and punctuation (commas,
colons,
>>>>>>> periods), such as "here is :" or "word , word"
>>>>>>> >> > - used sentence case for section titles instead
of headline case
>>>>>>> >> > - used abbreviations instead of the correct word
("ZK" instead
>>>>>>> of "ZooKeeper" being the biggest one here, but also "params"
instead of
>>>>>>> "parameters" is quite common)
>>>>>>> >> > - misspellings like "Zookeeper" instead of "ZooKeeper,
or
>>>>>>> "solr" instead of "Solr"
>>>>>>> >> > - config file names and parameter names/values
not in monospace
>>>>>>> >> > - lists of parameters are not properly formatted
(should not be
>>>>>>> in tables)
>>>>>>> >> >
>>>>>>> >> > These are all to make the Ref Guide as consistent,
cohesive,
>>>>>>> and easy to read as possible. It may be written by 30 people
but it
>>>>>>> shouldn't read like it is.
>>>>>>> >> >
>>>>>>> >> > Should I do all this while the commits are coming
through?
>>>>>>> Sure, but the reality is I can't. If we want to release the moment
someone
>>>>>>> proposes a release, then most of my find/replace list above needs
to go
>>>>>>> into precommit so these problems don't make it into the Guide
to begin
>>>>>>> with. (Which might be onerous since we'd all get stalled waiting
for
>>>>>>> someone to fix a typo...but really, precommit is meant in part
to find your
>>>>>>> typos so why should this be different?)
>>>>>>> >> >
>>>>>>> >> > It would always still need editorial review, however,
and
>>>>>>> that's not something we'll ever be able to fully automate. I'm
more than
>>>>>>> happy to have a little help there, but assume since people aren't
doing it
>>>>>>> today they don't have time, don't feel they have the skills,
or don't want
>>>>>>> to bother. Or maybe I just kill myself for a level of quality
no one else
>>>>>>> cares about...not sure I can stop doing it though if I'm the
RM.
>>>>>>> >> >
>>>>>>> >> > (as a side note on that though, if we do merge
the releases
>>>>>>> someday, then whoever RMs is going to have to wait for these
editorial
>>>>>>> processes to be completed or the vote may fail because the Ref
Guide reads
>>>>>>> like crap.)
>>>>>>> >> >
>>>>>>> >> > On Tue, Sep 4, 2018 at 11:33 AM jim ferenczi <
>>>>>>> jim.ferenczi@gmail.com> wrote:
>>>>>>> >> >>
>>>>>>> >> >> Thanks for explaining the situation Cassandra.
I was planning
>>>>>>> to build the first RC beginning of next week to give people a
week to
>>>>>>> discover blockers. I can certainly slow down things but I don't
think that
>>>>>>> the timing
>>>>>>> >> >> differs from other releases. I am not aware
of the operations
>>>>>>> that are required for the Ref guide release process but what
do you think
>>>>>>> of sharing the tasks with the RM ? We could even merge the two
releases and
>>>>>>> make the RM responsible of both if the process is documented.
 I'd be happy
>>>>>>> to experiment this for the 7.5 release if you want.
>>>>>>> >> >>
>>>>>>> >> >> Le mar. 4 sept. 2018 à 17:55, Cassandra Targett
<
>>>>>>> casstargett@gmail.com> a écrit :
>>>>>>> >> >>>
>>>>>>> >> >>> I'm not objecting per se, but I feel like
we used to propose
>>>>>>> a version and then give people a week before the branch was cut.
Maybe that
>>>>>>> was just RM choice? From a personal perspective, I much prefer
that model
>>>>>>> because the Ref Guide requires A LOT of my attention and my work
there
>>>>>>> kicks into high gear as soon as a release is proposed.
>>>>>>> >> >>>
>>>>>>> >> >>> Even though the artifact and Ref Guide
release processes are
>>>>>>> separate today, we want them to be a single process, so I need
to act as
>>>>>>> though your timeframe for the RC is the deadline for Ref Guide
edits to do
>>>>>>> an RC of the Ref Guide at the same time. That means I'm on your
timetable,
>>>>>>> no matter what else I may have promised to my bosses and colleagues.
It's
>>>>>>> stressful already to try to get it all done - I usually don't
finish
>>>>>>> everything I want to do - and adding the burden of having to
backport
>>>>>>> everything to 2 branches instead of 1 just makes it tedious as
well.
>>>>>>> >> >>>
>>>>>>> >> >>> Also, yesterday was a major holiday in
the US, and as of this
>>>>>>> moment it's not even noon on the East Coast, so there's a percentage
of the
>>>>>>> community who may not even have seen your proposal yet.
>>>>>>> >> >>>
>>>>>>> >> >>> I greatly appreciate that you've volunteered
to do the
>>>>>>> release and are energized to get it rolling, but is there a reason
an RC
>>>>>>> has to be done by the beginning of next week?
>>>>>>> >> >>>
>>>>>>> >> >>> On Tue, Sep 4, 2018 at 10:36 AM Joel Bernstein
<
>>>>>>> joelsolr@gmail.com> wrote:
>>>>>>> >> >>>>
>>>>>>> >> >>>> +1,
>>>>>>> >> >>>>
>>>>>>> >> >>>> I'll likely be adding some Solr RefGuide
changes later in
>>>>>>> the week to the 7.5 branch but I'll make sure they don't effect
the build.
>>>>>>> >> >>>>
>>>>>>> >> >>>>
>>>>>>> >> >>>> Joel Bernstein
>>>>>>> >> >>>> http://joelsolr.blogspot.com/
>>>>>>> >> >>>>
>>>>>>> >> >>>>
>>>>>>> >> >>>> On Tue, Sep 4, 2018 at 10:52 AM jim
ferenczi <
>>>>>>> jim.ferenczi@gmail.com> wrote:
>>>>>>> >> >>>>>
>>>>>>> >> >>>>> Thanks all,
>>>>>>> >> >>>>> since there are no objections I
am planning to cut the
>>>>>>> branch for 7.5 tomorrow. I'll build the first RC early next week
so there
>>>>>>> will be some room to merge important bug fixes later this week.
All
>>>>>>> blockers except SOLR-12727 seem to be merged/resolved, I'll watch
the
>>>>>>> remaining solr issue for updates.
>>>>>>> >> >>>>>
>>>>>>> >> >>>>> Le mar. 4 sept. 2018 à 10:21,
jim ferenczi <
>>>>>>> jim.ferenczi@gmail.com> a écrit :
>>>>>>> >> >>>>>>
>>>>>>> >> >>>>>> Sure Jan, this is a nice cleanup,
+1 to backport in 7x.
>>>>>>> >> >>>>>>
>>>>>>> >> >>>>>> Le mar. 4 sept. 2018 à 10:16,
Jan Høydahl <
>>>>>>> jan.asf@cominvent.com> a écrit :
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>> Jim, we have some release
process improvements in
>>>>>>> LUCENE-5143. Basically, we'll only have one KEYS file instead
of three plus
>>>>>>> those in the versioned folders that we have today. And the release
py
>>>>>>> script will start checking that the RM's key is present in the
KEYS file.
>>>>>>> Would you be ok with that being committed and you being the first
RM to use
>>>>>>> it for 7.5.0?
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>> --
>>>>>>> >> >>>>>>> Jan Høydahl, search solution
architect
>>>>>>> >> >>>>>>> Cominvent AS - www.cominvent.com
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>> 3. sep. 2018 kl. 10:42
skrev jim ferenczi <
>>>>>>> jim.ferenczi@gmail.com>:
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>> Hi all,
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>> 7.4 has been released two
months ago on June 29th and we
>>>>>>> have new features, enhancements and fixes that are not released
yet so I'd
>>>>>>> like to start working on releasing Lucene/Solr 7.5.0.
>>>>>>> >> >>>>>>> There's also a bad bug
with index sorting that deletes
>>>>>>> the wrong documents when delete by query is used:
>>>>>>> >> >>>>>>> https://issues.apache.org/jira/browse/LUCENE-8466
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>> I can create the 7.5 branch
later this week and build the
>>>>>>> first RC early next week if that works for everyone. Please let
me know if
>>>>>>> there are bug fixes that needs to be fixed in 7.5 and might not
be ready by
>>>>>>> then.
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>> Cheers,
>>>>>>> >> >>>>>>> Jim
>>>>>>> >> >>>>>>>
>>>>>>> >> >>>>>>>
>>>>>>> >>
>>>>>>> >>
>>>>>>> ---------------------------------------------------------------------
>>>>>>> >> To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
>>>>>>> >> For additional commands, e-mail: dev-help@lucene.apache.org
>>>>>>> >>
>>>>>>>
>>>>>>> ---------------------------------------------------------------------
>>>>>>> To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
>>>>>>> For additional commands, e-mail: dev-help@lucene.apache.org
>>>>>>>
>>>>>>>

Mime
View raw message