[Gluster-Maintainers] Update on experimental branch (was Re: Jenkins build is back to normal : experimental-periodic #21)

[Amar Tumballi]

On Wed, Jul 19, 2017 at 1:46 PM, Pranith Kumar Karampuri <
pkarampu at redhat.com> wrote:

>
>
> On Wed, Jul 19, 2017 at 1:32 PM, Atin Mukherjee <amukherj at redhat.com>
> wrote:
>
>>
>>
>> On Wed, Jul 19, 2017 at 11:36 AM, Amar Tumballi <atumball at redhat.com>
>> wrote:
>>
>>> Experimental branch is back to successful build now. So I think its a
>>> good time to update everyone about it.
>>>
>>> Currently the branch has changes for:
>>>
>>>
>>>    1. Added new on wire XDR changes.
>>>       - All is working as expected with newer protocol.
>>>       - So, if there is a change required for 4.0, I encourage everyone
>>>       to propose it and send a patch so we can test it out.
>>>    2. Changes in STACK_WIND/UNWIND to measure count and latency.
>>>    3. A separate way to dump metrics
>>>    - Similar to statedump, but only metrics, ie, a key and a value (as
>>>       int/float) as entries in the file.
>>>       - Some more improvements planned to provide functions in xlator_t
>>>       itself for all translator to dump their private metrics if any.
>>>       - I got a project going to draw these metrics in graphite/grafana
>>>       @ https://github.com/amarts/glustermetrics
>>>       4. Subdir mount
>>>       - Still waiting for authentication handling in
>>>       xlator/protocol/auth/addr to send it to master.
>>>       - Will take CLI changes after handling auth, thought for now is,
>>>       same auth.allow volume set option can be reused in early state.
>>>       5. New discover fop.
>>>       - Currently its implemented at xlator_t, defaults and all the
>>>       place.
>>>       - Saw major regressions with DHT / AFR by using it.
>>>       - Hence reverted the consumption of the discover fop for now.
>>>    6. Global Inode table.
>>>       - Most of the work was done and changes were present mainly in
>>>       inode/fd files of libglusterfs.
>>>       - Saw issues with regression tests, and some behaviour at the
>>>       moment, and have reverted the patch.
>>>
>>>
>>> Also it has one patch from Hari on tier add-brick/remove-brick.
>>>
>>>
>>> Considering I don't have more traffic on the branch yet, I am planning
>>> to branch out from master and rebasing all these changes on master again to
>>> stick with 3 months branch out timelines.
>>>
>>
>> I believe rebasing these changes to master will be done post 3.12
>> branching, right?
>>
>
>
When I am 'rebasing' its mostly `git rebase origin/master` on experiemental
branch and see if there are any conflicts. If none (or minor),  would
correct them and push out on the 'experimental' branch. This activity post
3.12 branch for sure.


> When you said rebase, you mean sending the patches on gerrit so that
> others can also review them right?
>
>

All the experimental branch patches are posted for review in gerrit, like
master branch or release branches. Till now I was mostly rebasing and
sending patches which were sent long back from me (before I left in 2013)
and getting the regression fixed, so pushed the patches quickly.

>From the re-branching post 3.12 branchout I will have at least 2 days for
any patch to get reviewed if its significant change.

If you are asking about me sending the changes to 'master' branch,  some of
it can be another week or 2, but some of these can be as high as 1 month or
so. For example, global inode change and discover change is not possible
without proper testing and significant changes, would be sent as close to
release 4.0 branch out if possible.

Regards,
Amar


>> Regards,
>>> Amar
>>>
>>> On Wed, Jul 19, 2017 at 11:07 AM, <jenkins at build.gluster.org> wrote:
>>>
>>>> See <https://build.gluster.org/job/experimental-periodic/21/disp
>>>> lay/redirect?page=changes>
>>>>
>>>>
>>>
>>>
>>> --
>>> Amar Tumballi (amarts)
>>>
>>> _______________________________________________
>>> maintainers mailing list
>>> maintainers at gluster.org
>>> http://lists.gluster.org/mailman/listinfo/maintainers
>>>
>>>
>>
>> _______________________________________________
>> maintainers mailing list
>> maintainers at gluster.org
>> http://lists.gluster.org/mailman/listinfo/maintainers
>>
>>
>
>
> --
> Pranith
>



-- 
Amar Tumballi (amarts)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gluster.org/pipermail/maintainers/attachments/20170719/3c9e4a62/attachment.html>