Merge lp:~dannf/charms/trusty/elasticsearch/unx86ify into lp:charms/trusty/elasticsearch

Proposed by dann frazier
Status: Merged
Merged at revision: 41
Proposed branch: lp:~dannf/charms/trusty/elasticsearch/unx86ify
Merge into: lp:charms/trusty/elasticsearch
Diff against target: 12 lines (+1/-1)
1 file modified
config.yaml (+1/-1)
To merge this branch: bzr merge lp:~dannf/charms/trusty/elasticsearch/unx86ify
Reviewer Review Type Date Requested Status
Marco Ceppi (community) Approve
Review via email: mp+246757@code.launchpad.net
To post a comment you must log in.
Revision history for this message
Marco Ceppi (marcoceppi) wrote :

LGTM

review: Approve
Revision history for this message
Michael Nelson (michael.nelson) wrote :

Hi Marco, Dann. I thought Dann and I had discussed this earlier, but I can't find the record (and old MP?). Generally, I'm not sure that updating the default options with specifics like this is a good idea - people with specific cases can override it as needed (and a mention in the README might be good). Specifically in this case, from what I remember, Dann you'd opened a bug upstream with elastico about this, and they'd explicitly said they weren't testing on this arch:

https://github.com/elastic/elasticsearch/issues/9299

"Closing this as we don;t want to provide packages for architecture which we are not testing. On these architectures Elasticsearch can still be installed from the zip or tar.gz distributions"

Revision history for this message
dann frazier (dannf) wrote :

On Mon, Nov 9, 2015 at 12:50 PM, Michael Nelson
<email address hidden> wrote:
> Hi Marco, Dann. I thought Dann and I had discussed this earlier, but I can't find the record (and old MP?).

I don't recall that - but it has been many months, so I wouldn't be
surprised. Apologies if I've forgotten - I definitely did not intend
to dismiss your feedback!

> Generally, I'm not sure that updating the default options with specifics like this is a good idea -
> people with specific cases can override it as needed (and a mention in the README might be good).

One of the goals of charms, aiui, is specifically to hide this type of
complexity. Leaving it up to a user to know that, because they happen
to be deploying on platform $foo, they need to specify config argument
$bar seems like magic--. If a charm can be made to *just work*,
independent of substrate, I'd like to achieve that. That said, I'm not
married to this specific implementation of making things just work -
it just seemed to fit the charm design the best, but I'm open to other
suggestions.

> Specifically in this case, from what I remember, Dann you'd opened a bug upstream with elastico about this, and they'd explicitly said they weren't testing on this arch:
>
> https://github.com/elastic/elasticsearch/issues/9299
>
> "Closing this as we don;t want to provide packages for architecture which we are not testing. On these architectures Elasticsearch can still be installed from the zip or tar.gz distributions"

Which I internalized as "the charm needs to workaround this for the
foreseeable future". That is, *I* know that this charm is functional
on arm64 (we demoed this at ODS last spring), and I know that there
users who *want* to use this charm to evaluate elasticsearch on arm64
hw. Of course, if there is a standard way to warn charm users they are
entering into unsupported land, I'd be all for adding such a warning
here.

  -dann

>
> --
> https://code.launchpad.net/~dannf/charms/trusty/elasticsearch/unx86ify/+merge/246757
> You are the owner of lp:~dannf/charms/trusty/elasticsearch/unx86ify.
>
> Launchpad-Message-Rationale: Owner
> Launchpad-Message-For: dannf
> Launchpad-Notification-Type: code-review
> Launchpad-Branch: ~dannf/charms/trusty/elasticsearch/unx86ify

Revision history for this message
Michael Nelson (michael.nelson) wrote :

On Tue, Nov 10, 2015 at 8:37 AM dann frazier <email address hidden>
wrote:

> On Mon, Nov 9, 2015 at 12:50 PM, Michael Nelson
> <email address hidden> wrote:
> > Hi Marco, Dann. I thought Dann and I had discussed this earlier, but I
> can't find the record (and old MP?).
>
> I don't recall that - but it has been many months, so I wouldn't be
> surprised. Apologies if I've forgotten - I definitely did not intend
> to dismiss your feedback!
>
> > Generally, I'm not sure that updating the default options with specifics
> like this is a good idea -
> > people with specific cases can override it as needed (and a mention in
> the README might be good).
>
> One of the goals of charms, aiui, is specifically to hide this type of
> complexity. Leaving it up to a user to know that, because they happen
> to be deploying on platform $foo, they need to specify config argument
> $bar seems like magic--. If a charm can be made to *just work*,
> independent of substrate, I'd like to achieve that. That said, I'm not
> married to this specific implementation of making things just work -
> it just seemed to fit the charm design the best, but I'm open to other
> suggestions.
>

Definitely agree - the charm should just work for all supported platforms...

>
> > Specifically in this case, from what I remember, Dann you'd opened a bug
> upstream with elastico about this, and they'd explicitly said they weren't
> testing on this arch:
> >
> > https://github.com/elastic/elasticsearch/issues/9299
> >
> > "Closing this as we don;t want to provide packages for architecture
> which we are not testing. On these architectures Elasticsearch can still be
> installed from the zip or tar.gz distributions"
>
> Which I internalized as "the charm needs to workaround this for the
> foreseeable future". That is, *I* know that this charm is functional
> on arm64 (we demoed this at ODS last spring), and I know that there
> users who *want* to use this charm to evaluate elasticsearch on arm64
> hw. Of course, if there is a standard way to warn charm users they are
> entering into unsupported land, I'd be all for adding such a warning
> here.
>

Yeah - that's what I'm uncertain about - and that's where a note in the
README for running and evaluating on unsupported arches would be easiest.
But I'm only -0, fine as is, just wanted to be explicit that elastico have
marked the above bug as wont-fix because they don't test on that platform.

Thanks Dann.

Preview Diff

[H/L] Next/Prev Comment, [J/K] Next/Prev File, [N/P] Next/Prev Hunk
1=== modified file 'config.yaml'
2--- config.yaml 2014-10-31 03:43:31 +0000
3+++ config.yaml 2015-01-16 18:01:31 +0000
4@@ -1,7 +1,7 @@
5 options:
6 apt-repository:
7 type: string
8- default: "deb http://packages.elasticsearch.org/elasticsearch/1.2/debian stable main"
9+ default: "deb [arch=amd64] http://packages.elasticsearch.org/elasticsearch/1.2/debian stable main"
10 description: |
11 A deb-line for the apt archive which contains the elasticsearch package.
12 This is necessary until elasticsearch gets into the debian/ubuntu archives.

Subscribers

People subscribed via source and target branches