Obsolete
Status Update
Comments
da...@gmail.com <da...@gmail.com> #2
[Empty comment from Monorail migration]
da...@gmail.com <da...@gmail.com> #3
[Empty comment from Monorail migration]
da...@gmail.com <da...@gmail.com> #4
[Empty comment from Monorail migration]
da...@gmail.com <da...@gmail.com> #5
As far as I remember, one of the reasons we documented ES as experimental was because none of us maintainers were/are using it in production.
I've heard that there are some users running ES in production now though. Would it be good to do a survey of the community to find out which versions (both of ES and of Gerrit) are actually being used in the wild? Is this something that the community managers can drive?
[Monorail components: Community]
I've heard that there are some users running ES in production now though. Would it be good to do a survey of the community to find out which versions (both of ES and of Gerrit) are actually being used in the wild? Is this something that the community managers can drive?
[Monorail components: Community]
al...@google.com <al...@google.com> #6
I think so yes; thanks. Now part of our near backlog.
There could be other Gerrit components to poll about; cf.https://crbug.com/gerrit/12211 .
There could be other Gerrit components to poll about; cf.
sv...@axis.com <sv...@axis.com> #7
Make a dedicated short survey for this to get this resolved quickly.
sv...@axis.com <sv...@axis.com> #8
[Empty comment from Monorail migration]
sv...@axis.com <sv...@axis.com> #9
Form proposal sent for review to some of us here.
Let me know if you cannot see it but would like to.
Once done reviewing, I can send the form/survey to the community mailing list.
-How much time to allow for collecting responses?
I'd propose by end of May, to give organizations a chance at figuring that out for themselves.
Let me know if you cannot see it but would like to.
Once done reviewing, I can send the form/survey to the community mailing list.
-How much time to allow for collecting responses?
I'd propose by end of May, to give organizations a chance at figuring that out for themselves.
sv...@axis.com <sv...@axis.com> #10
Looks good. I have just a couple of suggestions:
- Expand on the "both discontinued" statement to make it clearer what is meant. This is referring to the fact that those ES versions have reached EOL upstream, and support for them has been discontinued in Gerrit.
- I think it would be useful to also ask which version of Gerrit is being used. Then we can make a better judgement about whether it's safe to discontinue ES support in Gerrit's stable releases. Additionally it would help to decide in which stable branch we're going to remove the "experimental" caveat from the documentation.
- Expand on the "both discontinued" statement to make it clearer what is meant. This is referring to the fact that those ES versions have reached EOL upstream, and support for them has been discontinued in Gerrit.
- I think it would be useful to also ask which version of Gerrit is being used. Then we can make a better judgement about whether it's safe to discontinue ES support in Gerrit's stable releases. Additionally it would help to decide in which stable branch we're going to remove the "experimental" caveat from the documentation.
sv...@axis.com <sv...@axis.com> #11
Thanks David.
- I added a "Discontinued Elasticsearch versions" section from your first suggestion above, with a few more words and upstream reference.
- I added 2 questions based on your second suggestion; good point as well.
I also added a trailing note about the survey end date (May end).
And I added Edwin's email suggestions to the survey intro.
I altered some questions to fit the "no Elasticsearch use" case in; thanks Edwin.
Can one of the form reviewer(s) test it, or I can send it as is; let me know.
- I added a "Discontinued Elasticsearch versions" section from your first suggestion above, with a few more words and upstream reference.
- I added 2 questions based on your second suggestion; good point as well.
I also added a trailing note about the survey end date (May end).
And I added Edwin's email suggestions to the survey intro.
I altered some questions to fit the "no Elasticsearch use" case in; thanks Edwin.
Can one of the form reviewer(s) test it, or I can send it as is; let me know.
sv...@axis.com <sv...@axis.com> #12
The form looks good now. I think it's OK to send it, but let's wait for comments from anyone else.
hi...@google.com <hi...@google.com> #13
I just realized that I forgot David (O.), so I just invited him to the form.
I'm done receiving go-aheads from everyone else by now.
I'm then planning to post the survey on Monday morning my time.
-This is so that I
1. can check David-O's comments if any, and
2. skip the lower email attendance /week-end.
I'm done receiving go-aheads from everyone else by now.
I'm then planning to post the survey on Monday morning my time.
-This is so that I
1. can check David-O's comments if any, and
2. skip the lower email attendance /week-end.
sv...@axis.com <sv...@axis.com> #14
> I just realized that I forgot David (O.), so I just invited him to the form.
Thanks, looks good. I have not found a way how to comment on the
Form document, answering here:
Can we add additional question if ES not used currently:
Are your organization considering/planning to switch to using Elasticsearch in near future?
Possible further question if asnwer to the above question is Yes/No:
Yes:
What specific version?
No:
Why not?
Thanks, looks good. I have not found a way how to comment on the
Form document, answering here:
Can we add additional question if ES not used currently:
Are your organization considering/planning to switch to using Elasticsearch in near future?
Possible further question if asnwer to the above question is Yes/No:
Yes:
What specific version?
No:
Why not?
da...@gmail.com <da...@gmail.com> #15
Thanks David. -Done.
Survey form is posted to the list now; awaiting responses...
We will monitor upcoming responses until May end.
I'm setting this Issue to Started under my name until then.
Responses should teach us how to tackle versions and beta-ness of ES in Gerrit.
Survey form is posted to the list now; awaiting responses...
We will monitor upcoming responses until May end.
I'm setting this Issue to Started under my name until then.
Responses should teach us how to tackle versions and beta-ness of ES in Gerrit.
da...@gmail.com <da...@gmail.com> #16
Survey is now over thus no longer accepting responses; results:
1. There were 10 responses in total.
2. Only 1 respondent is formally using Elasticsearch in Gerrit (2.16.7), and it is version 6.3.2.
3. One respondent is planning to use version 7.6 (or so) this year.
4. Another respondent might use the latest version this year.
5. The question about which Gerrit is used spread across all versions, with 3.1 having more respondents (2 out of 10).
So the other respondents spread evenly from <=2.15 to 3.2/master.
But that question was likely misunderstood, or it had an issue in the form, as it was meant to relate to Elasticsearch.
-No big deal anyway.
In light of these limited survey results,
6. we could consider making version 6 non-experimental in the Gerrit documentation.
7. As for version 7, it could remain 'experimental' until proven otherwise (through potential production later this year).
-More thoughts?
1. There were 10 responses in total.
2. Only 1 respondent is formally using Elasticsearch in Gerrit (2.16.7), and it is version 6.3.2.
3. One respondent is planning to use version 7.6 (or so) this year.
4. Another respondent might use the latest version this year.
5. The question about which Gerrit is used spread across all versions, with 3.1 having more respondents (2 out of 10).
So the other respondents spread evenly from <=2.15 to 3.2/master.
But that question was likely misunderstood, or it had an issue in the form, as it was meant to relate to Elasticsearch.
-No big deal anyway.
In light of these limited survey results,
6. we could consider making version 6 non-experimental in the Gerrit documentation.
7. As for version 7, it could remain 'experimental' until proven otherwise (through potential production later this year).
-More thoughts?
sv...@axis.com <sv...@axis.com> #17
sv...@axis.com <sv...@axis.com> #18
Having only 1 response isn't enough IMHO to be representative. Also, are we saying that we support ES in production for a EOL release?
I would read the results as "really nobody except one uses it".
I would focus in making it production-ready for v3.3 and on ES 7, using our E2E suite and testing it in a production-like environment.
WDYT?
I would read the results as "really nobody except one uses it".
I would focus in making it production-ready for v3.3 and on ES 7, using our E2E suite and testing it in a production-like environment.
WDYT?
al...@google.com <al...@google.com> #19
> Having only 1 response isn't enough IMHO to be representative.
I agree(d) to that; indeed.
> Also, are we saying that we support ES in production for a EOL release?
Yes, through the proposed [1], because this is the only community data we have so far:
[1]https://gerrit-review.googlesource.com/c/homepage/+/269993
Let's comment further in [1]'s thread.
Note that we are still documenting EOL versions within that page.
> I would read the results as "really nobody except one uses it".
True.
> I would focus in making it production-ready for v3.3 and on ES 7, using our E2E suite and testing it in a production-like environment.
We can do that yes, as well as consider initializing the documentation for now ([1] above).
I agree(d) to that; indeed.
> Also, are we saying that we support ES in production for a EOL release?
Yes, through the proposed [1], because this is the only community data we have so far:
[1]
Let's comment further in [1]'s thread.
Note that we are still documenting EOL versions within that page.
> I would read the results as "really nobody except one uses it".
True.
> I would focus in making it production-ready for v3.3 and on ES 7, using our E2E suite and testing it in a production-like environment.
We can do that yes, as well as consider initializing the documentation for now ([1] above).
sv...@axis.com <sv...@axis.com> #20
Use the E2E suite with ES 7 on 3.2 would be useful as would testing offline reindex required by noteDb migration in 2.16.
sv...@axis.com <sv...@axis.com> #21
> In fact I'd rather document 6.3.2 as known to be reliable enough with 2.16.7, to start with.
I don't think it helps much to document this, since not only is Gerrit 2.16.x EOL but also ES 6.3.2 is EOL and support was discontinued in Gerrit 2.16.18.
I don't think it helps much to document this, since not only is Gerrit 2.16.x EOL but also ES 6.3.2 is EOL and support was discontinued in Gerrit 2.16.18.
Description
As part of the assessment if we should do this, the ESC looked into the plugin from different dimensions:
1) UI/UX (Ben)
2) Usage (Luca)
3) Applicability (Alice)
4) Code quality (Patrick).
This bug tracks the outcome of looking over the code of the plugin and identify gaps that should be fixed before it becomes a core plugin.
The plugin doesn't have a lot of backend code (good). The style and usage of APIs is in-line with what we would expect in Gerrit core. There are integration tests (good) that test configuration and usage of the plugin (calling it via extension points in core).
1) Add JavaDoc to classes. One-liners are OK, but intent should be explained, so that casual readers understand how it fits together.
2) ReviewersIT uses Thread.sleep to wait out calls from Gerrit core / processing. Core is calling plugins synchronously, so it should be possible to get the test written without thread waiting. The plugin internally uses a WorkQueue, which makes processing async, but we can bind that to a directExecutor() in the test to make it sync and remove the Thread waiting.
3) The "test" prefixes should be removed from methods to match style in core (accountGroupResolve instead of testAccountGroupResolve)
4) Add tests for special case changes (private, WiP)
5) Add tests for change visibility (configured to add a user as reviewer that can't see the change)