We already do “bulk/batch” release voting, as every vote covers several
artifacts (geode, geode-native, geode-benchmarks, geode-examples). I don’t
interpret the ASF rules as setting any limit on the number of artifacts that
can be approved in one vote.
It sounds like the rough consensus from
I thought that was the case…
Then maybe every release has to be voted on, no bulk/batch allowed, as every
release has to be verified separately.
Which, just means that bulk announcements is the only thing that makes sense…
From: Owen Nichols
Date: Friday, December 3, 2021 at 10:39 AM
To:
I wonder if voting on patch versions makes sense.
Releasing (even patch releases) is an Act of the Apache Software Foundation,
and as such there is a legal obligation to follow ASF voting rules for every
release. Technically the purpose of the vote is for the Geode PMC to certify
that the
I wonder if voting on patch versions makes sense. As we should never be
breaking any existing features and essentially there should be sufficient
testing on the fixes to confirm that they resolve the issues. There should also
be no changes to APIs, as those changes should be included in a new
It doesn’t really make a lot of sense to release a fix in a 1.12 patch
separately without later patches. Doing so would send the confusing message
that users need to downgrade to get the fix (or if they upgrade they will lose
it).
So, whether we vote separately or not, we should still probably
The upgrade is complete, so everything should be working normally.
On Thu, Dec 2, 2021 at 2:05 PM Sean Goller wrote:
> Due to some memory pressure we need to increase the size of some nodes in
> the ci infrastructure. Due to this there may be some instability with
> concourse, but we don't
I would hope that all RCs had been checked thoroughly before any release VOTE
occurs.
From: Donal Evans
Date: Thursday, December 2, 2021 at 11:47 AM
To: dev@geode.apache.org
Subject: Re: [DISCUSS] batch patch release voting
The thing I wonder about with this decision is what happens if there
Due to some memory pressure we need to increase the size of some nodes in
the ci infrastructure. Due to this there may be some instability with
concourse, but we don't believe there will be any. This is just a heads up
in case something happens.
-Sean.
The thing I wonder about with this decision is what happens if there are no
problems with two of the releases, but one of them has some showstopping issue
that needs to be addressed.
Given that not every fix that's going into 1.14.1 is going into 1.12.6, it's
possible that an unrelated issue
+1 for batching them.
Is it implied with a +1 vote that all the releases have been checked or can you
only vote on a single release if that's all you checked? We can define this
when we send out the vote email.
-Original Message-
From: Robert Houghton
Sent: Thursday, December 2,
In the case where its one change being backported all over, I say batch up the
vote.
From: Owen Nichols
Date: Wednesday, December 1, 2021 at 11:46 AM
To: dev@geode.apache.org
Subject: [DISCUSS] batch patch release voting
Geode's N-2 support policy can lead to "waves" of patch releases. For
Hi everyone,
Sorry, I didn't organize the December community meeting that should have taken
place today. Given, it's the holiday season coming up, I wonder if it makes
sense to not bother rescheduling and instead start looking at our January
meeting.
Does anyone have a topic they'd like to
Related but different question: how many active oplogs do you normally see at
one time? You may want to adjust the max-oplog-size if the default of 1 GB is
too small.
On Dec 2, 2021, at 1:11 AM, Jakov Varenina
mailto:jakov.varen...@est.tech>> wrote:
Hi Dan,
We forget to mention that we
Hi Dan,
It's clear that supporting this case is tricky, both technically and in terms
of security. However, luckily that's not the goal here.
Thing is what we've observed is during some scenario (probably while using a
proxy, like envoy), the client receives a response from locators which is
Hi Dan,
We forget to mention that we actually configure off-heap for the
regions, so cache entry values are stored outside the heap memory. Only
Oplog objects that are not compacted and that have .crf file are
referencing the live entries from the cache. These Oplog objects are not
stored in
15 matches
Mail list logo
