rbd-mirror related jewel backports

rbd-mirror related jewel backports

[Loic Dachary]

Hi Mykola,

There are seven rbd-mirror related backports which do not cherry-pick cleanly. I've not looked at each of them yet and I would very much appreciate your advice about the best way to proceed.

http://tracker.ceph.com/issues/16511
http://tracker.ceph.com/issues/16512
http://tracker.ceph.com/issues/16658
http://tracker.ceph.com/issues/16701
http://tracker.ceph.com/issues/16747
http://tracker.ceph.com/issues/16902
http://tracker.ceph.com/issues/16978

Maybe they should be cherry-picked in a specific order in order to avoid conflicts ? For instance, maybe one of the pending backports that you can see at

https://github.com/ceph/ceph/pulls?q=is%3Aopen+is%3Apr+milestone%3Ajewel+label%3Arbd

would avoid some of the conflicts if merged ? I'm thinking specifically

https://github.com/ceph/ceph/pull/10646
https://github.com/ceph/ceph/pull/10644

What I'm looking for, ideally, is a way to order the cherry-picks to avoid any conflict. That may involve backporting a few additional pull requests (for instance when the type of an argument changes from int64_t to uint64_t, that creates a lot of potential conflicts and backporting that change could help, even if it is not required to fix a given bug.

There are three backports that also do not apply because of a conflict but do not seem to be related to rbd-mirror. Unless you advise otherwise, I'll attempt to resolve these individually and assume there is complex interaction with other backports.

http://tracker.ceph.com/issues/16950
http://tracker.ceph.com/issues/16904
http://tracker.ceph.com/issues/16735

Thanks in advance for your guidance !

-- 
Loïc Dachary, Artisan Logiciel Libre

Re: rbd-mirror related jewel backports

[Jason Dillaman]

Loic,

If you want, I can create the backport PRs for those tickets.


On Wed, Aug 10, 2016 at 5:07 AM, Loic Dachary <loic@dachary.org> wrote:
> Hi Mykola,
>
> There are seven rbd-mirror related backports which do not cherry-pick cleanly. I've not looked at each of them yet and I would very much appreciate your advice about the best way to proceed.
>
> http://tracker.ceph.com/issues/16511
> http://tracker.ceph.com/issues/16512
> http://tracker.ceph.com/issues/16658
> http://tracker.ceph.com/issues/16701
> http://tracker.ceph.com/issues/16747
> http://tracker.ceph.com/issues/16902
> http://tracker.ceph.com/issues/16978
>
> Maybe they should be cherry-picked in a specific order in order to avoid conflicts ? For instance, maybe one of the pending backports that you can see at
>
> https://github.com/ceph/ceph/pulls?q=is%3Aopen+is%3Apr+milestone%3Ajewel+label%3Arbd
>
> would avoid some of the conflicts if merged ? I'm thinking specifically
>
> https://github.com/ceph/ceph/pull/10646
> https://github.com/ceph/ceph/pull/10644
>
> What I'm looking for, ideally, is a way to order the cherry-picks to avoid any conflict. That may involve backporting a few additional pull requests (for instance when the type of an argument changes from int64_t to uint64_t, that creates a lot of potential conflicts and backporting that change could help, even if it is not required to fix a given bug.
>
> There are three backports that also do not apply because of a conflict but do not seem to be related to rbd-mirror. Unless you advise otherwise, I'll attempt to resolve these individually and assume there is complex interaction with other backports.
>
> http://tracker.ceph.com/issues/16950
> http://tracker.ceph.com/issues/16904
> http://tracker.ceph.com/issues/16735
>
> Thanks in advance for your guidance !
>
> --
> Loïc Dachary, Artisan Logiciel Libre
>
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html



-- 
Jason

Re: rbd-mirror related jewel backports

[Loic Dachary]

On 10/08/2016 22:14, Jason Dillaman wrote:
> Loic,
> 
> If you want, I can create the backport PRs for those tickets.

That would be great. Alternatively, if that's less time consuming for you, just suggest a backport order and I'll figure it out.

Cheers

> 
> 
> On Wed, Aug 10, 2016 at 5:07 AM, Loic Dachary <loic@dachary.org> wrote:
>> Hi Mykola,
>>
>> There are seven rbd-mirror related backports which do not cherry-pick cleanly. I've not looked at each of them yet and I would very much appreciate your advice about the best way to proceed.
>>
>> http://tracker.ceph.com/issues/16511
>> http://tracker.ceph.com/issues/16512
>> http://tracker.ceph.com/issues/16658
>> http://tracker.ceph.com/issues/16701
>> http://tracker.ceph.com/issues/16747
>> http://tracker.ceph.com/issues/16902
>> http://tracker.ceph.com/issues/16978
>>
>> Maybe they should be cherry-picked in a specific order in order to avoid conflicts ? For instance, maybe one of the pending backports that you can see at
>>
>> https://github.com/ceph/ceph/pulls?q=is%3Aopen+is%3Apr+milestone%3Ajewel+label%3Arbd
>>
>> would avoid some of the conflicts if merged ? I'm thinking specifically
>>
>> https://github.com/ceph/ceph/pull/10646
>> https://github.com/ceph/ceph/pull/10644
>>
>> What I'm looking for, ideally, is a way to order the cherry-picks to avoid any conflict. That may involve backporting a few additional pull requests (for instance when the type of an argument changes from int64_t to uint64_t, that creates a lot of potential conflicts and backporting that change could help, even if it is not required to fix a given bug.
>>
>> There are three backports that also do not apply because of a conflict but do not seem to be related to rbd-mirror. Unless you advise otherwise, I'll attempt to resolve these individually and assume there is complex interaction with other backports.
>>
>> http://tracker.ceph.com/issues/16950
>> http://tracker.ceph.com/issues/16904
>> http://tracker.ceph.com/issues/16735
>>
>> Thanks in advance for your guidance !
>>
>> --
>> Loïc Dachary, Artisan Logiciel Libre
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> 
> 

-- 
Loïc Dachary, Artisan Logiciel Libre

Re: rbd-mirror related jewel backports

[Jason Dillaman]

No worries.

My usual routine just involves pulling up master and jewel branch
merge logs for all RBD-related subdirectories in two terminals (for PR
merge timeline reference), pulling up the original PR to backport, and
then just cherry-pick until a conflict is hit. When that happens, I
locate which previous PR made the conflicting change and cherry-pick
it into the same backport branch.

With this technique, you'll end up with PRs with the same
cherry-picked commits.  You'll be able to cleanly merge them into a
testing branch, but when it comes time to merge into the jewel branch
you'll need to rebase some PRs to automatically strip out the
duplicate cherry-picks.

On Wed, Aug 10, 2016 at 5:00 PM, Loic Dachary <loic@dachary.org> wrote:
>
>
> On 10/08/2016 22:14, Jason Dillaman wrote:
>> Loic,
>>
>> If you want, I can create the backport PRs for those tickets.
>
> That would be great. Alternatively, if that's less time consuming for you, just suggest a backport order and I'll figure it out.
>
> Cheers
>
>>
>>
>> On Wed, Aug 10, 2016 at 5:07 AM, Loic Dachary <loic@dachary.org> wrote:
>>> Hi Mykola,
>>>
>>> There are seven rbd-mirror related backports which do not cherry-pick cleanly. I've not looked at each of them yet and I would very much appreciate your advice about the best way to proceed.
>>>
>>> http://tracker.ceph.com/issues/16511
>>> http://tracker.ceph.com/issues/16512
>>> http://tracker.ceph.com/issues/16658
>>> http://tracker.ceph.com/issues/16701
>>> http://tracker.ceph.com/issues/16747
>>> http://tracker.ceph.com/issues/16902
>>> http://tracker.ceph.com/issues/16978
>>>
>>> Maybe they should be cherry-picked in a specific order in order to avoid conflicts ? For instance, maybe one of the pending backports that you can see at
>>>
>>> https://github.com/ceph/ceph/pulls?q=is%3Aopen+is%3Apr+milestone%3Ajewel+label%3Arbd
>>>
>>> would avoid some of the conflicts if merged ? I'm thinking specifically
>>>
>>> https://github.com/ceph/ceph/pull/10646
>>> https://github.com/ceph/ceph/pull/10644
>>>
>>> What I'm looking for, ideally, is a way to order the cherry-picks to avoid any conflict. That may involve backporting a few additional pull requests (for instance when the type of an argument changes from int64_t to uint64_t, that creates a lot of potential conflicts and backporting that change could help, even if it is not required to fix a given bug.
>>>
>>> There are three backports that also do not apply because of a conflict but do not seem to be related to rbd-mirror. Unless you advise otherwise, I'll attempt to resolve these individually and assume there is complex interaction with other backports.
>>>
>>> http://tracker.ceph.com/issues/16950
>>> http://tracker.ceph.com/issues/16904
>>> http://tracker.ceph.com/issues/16735
>>>
>>> Thanks in advance for your guidance !
>>>
>>> --
>>> Loïc Dachary, Artisan Logiciel Libre
>>>
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
>>> the body of a message to majordomo@vger.kernel.org
>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>>
>>
>
> --
> Loïc Dachary, Artisan Logiciel Libre



-- 
Jason

Re: rbd-mirror related jewel backports

[Loic Dachary]

I can try to do that, it is simple enough. Because I'm not very familiar with the code base, it is entirely possible that I end up cherry-picking a lot more than what is reasonable when doing that. It is sometime non trivial to figure out when to stop. Who knows: I may end up cherry-picking *everything* from master :-) I'll call for help if it drifts too much.

Cheers

On 11/08/2016 15:22, Jason Dillaman wrote:
> No worries.
> 
> My usual routine just involves pulling up master and jewel branch
> merge logs for all RBD-related subdirectories in two terminals (for PR
> merge timeline reference), pulling up the original PR to backport, and
> then just cherry-pick until a conflict is hit. When that happens, I
> locate which previous PR made the conflicting change and cherry-pick
> it into the same backport branch.
> 
> With this technique, you'll end up with PRs with the same
> cherry-picked commits.  You'll be able to cleanly merge them into a
> testing branch, but when it comes time to merge into the jewel branch
> you'll need to rebase some PRs to automatically strip out the
> duplicate cherry-picks.
> 
> On Wed, Aug 10, 2016 at 5:00 PM, Loic Dachary <loic@dachary.org> wrote:
>>
>>
>> On 10/08/2016 22:14, Jason Dillaman wrote:
>>> Loic,
>>>
>>> If you want, I can create the backport PRs for those tickets.
>>
>> That would be great. Alternatively, if that's less time consuming for you, just suggest a backport order and I'll figure it out.
>>
>> Cheers
>>
>>>
>>>
>>> On Wed, Aug 10, 2016 at 5:07 AM, Loic Dachary <loic@dachary.org> wrote:
>>>> Hi Mykola,
>>>>
>>>> There are seven rbd-mirror related backports which do not cherry-pick cleanly. I've not looked at each of them yet and I would very much appreciate your advice about the best way to proceed.
>>>>
>>>> http://tracker.ceph.com/issues/16511
>>>> http://tracker.ceph.com/issues/16512
>>>> http://tracker.ceph.com/issues/16658
>>>> http://tracker.ceph.com/issues/16701
>>>> http://tracker.ceph.com/issues/16747
>>>> http://tracker.ceph.com/issues/16902
>>>> http://tracker.ceph.com/issues/16978
>>>>
>>>> Maybe they should be cherry-picked in a specific order in order to avoid conflicts ? For instance, maybe one of the pending backports that you can see at
>>>>
>>>> https://github.com/ceph/ceph/pulls?q=is%3Aopen+is%3Apr+milestone%3Ajewel+label%3Arbd
>>>>
>>>> would avoid some of the conflicts if merged ? I'm thinking specifically
>>>>
>>>> https://github.com/ceph/ceph/pull/10646
>>>> https://github.com/ceph/ceph/pull/10644
>>>>
>>>> What I'm looking for, ideally, is a way to order the cherry-picks to avoid any conflict. That may involve backporting a few additional pull requests (for instance when the type of an argument changes from int64_t to uint64_t, that creates a lot of potential conflicts and backporting that change could help, even if it is not required to fix a given bug.
>>>>
>>>> There are three backports that also do not apply because of a conflict but do not seem to be related to rbd-mirror. Unless you advise otherwise, I'll attempt to resolve these individually and assume there is complex interaction with other backports.
>>>>
>>>> http://tracker.ceph.com/issues/16950
>>>> http://tracker.ceph.com/issues/16904
>>>> http://tracker.ceph.com/issues/16735
>>>>
>>>> Thanks in advance for your guidance !
>>>>
>>>> --
>>>> Loïc Dachary, Artisan Logiciel Libre
>>>>
>>>> --
>>>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
>>>> the body of a message to majordomo@vger.kernel.org
>>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>>
>>>
>>>
>>
>> --
>> Loïc Dachary, Artisan Logiciel Libre
> 
> 
> 

-- 
Loïc Dachary, Artisan Logiciel Libre

Re: rbd-mirror related jewel backports

[Jason Dillaman]

OK, that's fine with me.  The only ones left on your original list are
(in order):

http://tracker.ceph.com/issues/16902 (master 10484) (Tue, 2 Aug 2016
20:14:54 +0300)
http://tracker.ceph.com/issues/16950 (master 10574) (Sun, 7 Aug 2016
13:33:45 +0300)
http://tracker.ceph.com/issues/16978 (master 10613) (Tue, 9 Aug 2016
16:18:49 +0300)

On Thu, Aug 11, 2016 at 9:58 AM, Loic Dachary <loic@dachary.org> wrote:
> I can try to do that, it is simple enough. Because I'm not very familiar with the code base, it is entirely possible that I end up cherry-picking a lot more than what is reasonable when doing that. It is sometime non trivial to figure out when to stop. Who knows: I may end up cherry-picking *everything* from master :-) I'll call for help if it drifts too much.
>
> Cheers
>
> On 11/08/2016 15:22, Jason Dillaman wrote:
>> No worries.
>>
>> My usual routine just involves pulling up master and jewel branch
>> merge logs for all RBD-related subdirectories in two terminals (for PR
>> merge timeline reference), pulling up the original PR to backport, and
>> then just cherry-pick until a conflict is hit. When that happens, I
>> locate which previous PR made the conflicting change and cherry-pick
>> it into the same backport branch.
>>
>> With this technique, you'll end up with PRs with the same
>> cherry-picked commits.  You'll be able to cleanly merge them into a
>> testing branch, but when it comes time to merge into the jewel branch
>> you'll need to rebase some PRs to automatically strip out the
>> duplicate cherry-picks.
>>
>> On Wed, Aug 10, 2016 at 5:00 PM, Loic Dachary <loic@dachary.org> wrote:
>>>
>>>
>>> On 10/08/2016 22:14, Jason Dillaman wrote:
>>>> Loic,
>>>>
>>>> If you want, I can create the backport PRs for those tickets.
>>>
>>> That would be great. Alternatively, if that's less time consuming for you, just suggest a backport order and I'll figure it out.
>>>
>>> Cheers
>>>
>>>>
>>>>
>>>> On Wed, Aug 10, 2016 at 5:07 AM, Loic Dachary <loic@dachary.org> wrote:
>>>>> Hi Mykola,
>>>>>
>>>>> There are seven rbd-mirror related backports which do not cherry-pick cleanly. I've not looked at each of them yet and I would very much appreciate your advice about the best way to proceed.
>>>>>
>>>>> http://tracker.ceph.com/issues/16511
>>>>> http://tracker.ceph.com/issues/16512
>>>>> http://tracker.ceph.com/issues/16658
>>>>> http://tracker.ceph.com/issues/16701
>>>>> http://tracker.ceph.com/issues/16747
>>>>> http://tracker.ceph.com/issues/16902
>>>>> http://tracker.ceph.com/issues/16978
>>>>>
>>>>> Maybe they should be cherry-picked in a specific order in order to avoid conflicts ? For instance, maybe one of the pending backports that you can see at
>>>>>
>>>>> https://github.com/ceph/ceph/pulls?q=is%3Aopen+is%3Apr+milestone%3Ajewel+label%3Arbd
>>>>>
>>>>> would avoid some of the conflicts if merged ? I'm thinking specifically
>>>>>
>>>>> https://github.com/ceph/ceph/pull/10646
>>>>> https://github.com/ceph/ceph/pull/10644
>>>>>
>>>>> What I'm looking for, ideally, is a way to order the cherry-picks to avoid any conflict. That may involve backporting a few additional pull requests (for instance when the type of an argument changes from int64_t to uint64_t, that creates a lot of potential conflicts and backporting that change could help, even if it is not required to fix a given bug.
>>>>>
>>>>> There are three backports that also do not apply because of a conflict but do not seem to be related to rbd-mirror. Unless you advise otherwise, I'll attempt to resolve these individually and assume there is complex interaction with other backports.
>>>>>
>>>>> http://tracker.ceph.com/issues/16950
>>>>> http://tracker.ceph.com/issues/16904
>>>>> http://tracker.ceph.com/issues/16735
>>>>>
>>>>> Thanks in advance for your guidance !
>>>>>
>>>>> --
>>>>> Loïc Dachary, Artisan Logiciel Libre
>>>>>
>>>>> --
>>>>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
>>>>> the body of a message to majordomo@vger.kernel.org
>>>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>>>
>>>>
>>>>
>>>
>>> --
>>> Loïc Dachary, Artisan Logiciel Libre
>>
>>
>>
>
> --
> Loïc Dachary, Artisan Logiciel Libre



-- 
Jason

Re: rbd-mirror related jewel backports

[Loic Dachary]

Hi Jason,

On 11/08/2016 17:01, Jason Dillaman wrote:
> OK, that's fine with me.  The only ones left on your original list are
> (in order):

Excellent, that makes our work a lot easier :-) I'll start testing https://github.com/ceph/ceph/pull/10678 and once merged it will clear 6 backports.

> 
> http://tracker.ceph.com/issues/16902 (master 10484) (Tue, 2 Aug 2016
> 20:14:54 +0300)
> http://tracker.ceph.com/issues/16950 (master 10574) (Sun, 7 Aug 2016
> 13:33:45 +0300)
> http://tracker.ceph.com/issues/16978 (master 10613) (Tue, 9 Aug 2016
> 16:18:49 +0300)
> 
> On Thu, Aug 11, 2016 at 9:58 AM, Loic Dachary <loic@dachary.org> wrote:
>> I can try to do that, it is simple enough. Because I'm not very familiar with the code base, it is entirely possible that I end up cherry-picking a lot more than what is reasonable when doing that. It is sometime non trivial to figure out when to stop. Who knows: I may end up cherry-picking *everything* from master :-) I'll call for help if it drifts too much.
>>
>> Cheers
>>
>> On 11/08/2016 15:22, Jason Dillaman wrote:
>>> No worries.
>>>
>>> My usual routine just involves pulling up master and jewel branch
>>> merge logs for all RBD-related subdirectories in two terminals (for PR
>>> merge timeline reference), pulling up the original PR to backport, and
>>> then just cherry-pick until a conflict is hit. When that happens, I
>>> locate which previous PR made the conflicting change and cherry-pick
>>> it into the same backport branch.
>>>
>>> With this technique, you'll end up with PRs with the same
>>> cherry-picked commits.  You'll be able to cleanly merge them into a
>>> testing branch, but when it comes time to merge into the jewel branch
>>> you'll need to rebase some PRs to automatically strip out the
>>> duplicate cherry-picks.
>>>
>>> On Wed, Aug 10, 2016 at 5:00 PM, Loic Dachary <loic@dachary.org> wrote:
>>>>
>>>>
>>>> On 10/08/2016 22:14, Jason Dillaman wrote:
>>>>> Loic,
>>>>>
>>>>> If you want, I can create the backport PRs for those tickets.
>>>>
>>>> That would be great. Alternatively, if that's less time consuming for you, just suggest a backport order and I'll figure it out.
>>>>
>>>> Cheers
>>>>
>>>>>
>>>>>
>>>>> On Wed, Aug 10, 2016 at 5:07 AM, Loic Dachary <loic@dachary.org> wrote:
>>>>>> Hi Mykola,
>>>>>>
>>>>>> There are seven rbd-mirror related backports which do not cherry-pick cleanly. I've not looked at each of them yet and I would very much appreciate your advice about the best way to proceed.
>>>>>>
>>>>>> http://tracker.ceph.com/issues/16511
>>>>>> http://tracker.ceph.com/issues/16512
>>>>>> http://tracker.ceph.com/issues/16658
>>>>>> http://tracker.ceph.com/issues/16701
>>>>>> http://tracker.ceph.com/issues/16747
>>>>>> http://tracker.ceph.com/issues/16902
>>>>>> http://tracker.ceph.com/issues/16978
>>>>>>
>>>>>> Maybe they should be cherry-picked in a specific order in order to avoid conflicts ? For instance, maybe one of the pending backports that you can see at
>>>>>>
>>>>>> https://github.com/ceph/ceph/pulls?q=is%3Aopen+is%3Apr+milestone%3Ajewel+label%3Arbd
>>>>>>
>>>>>> would avoid some of the conflicts if merged ? I'm thinking specifically
>>>>>>
>>>>>> https://github.com/ceph/ceph/pull/10646
>>>>>> https://github.com/ceph/ceph/pull/10644
>>>>>>
>>>>>> What I'm looking for, ideally, is a way to order the cherry-picks to avoid any conflict. That may involve backporting a few additional pull requests (for instance when the type of an argument changes from int64_t to uint64_t, that creates a lot of potential conflicts and backporting that change could help, even if it is not required to fix a given bug.
>>>>>>
>>>>>> There are three backports that also do not apply because of a conflict but do not seem to be related to rbd-mirror. Unless you advise otherwise, I'll attempt to resolve these individually and assume there is complex interaction with other backports.
>>>>>>
>>>>>> http://tracker.ceph.com/issues/16950
>>>>>> http://tracker.ceph.com/issues/16904
>>>>>> http://tracker.ceph.com/issues/16735
>>>>>>
>>>>>> Thanks in advance for your guidance !
>>>>>>
>>>>>> --
>>>>>> Loïc Dachary, Artisan Logiciel Libre
>>>>>>
>>>>>> --
>>>>>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
>>>>>> the body of a message to majordomo@vger.kernel.org
>>>>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>>>>
>>>>>
>>>>>
>>>>
>>>> --
>>>> Loïc Dachary, Artisan Logiciel Libre
>>>
>>>
>>>
>>
>> --
>> Loïc Dachary, Artisan Logiciel Libre
> 
> 
> 

-- 
Loïc Dachary, Artisan Logiciel Libre