Thanks for the reply, really helpful!
We will have a try for the different methods deployment and will let you know if there are some further questions.
On Tue, 12 Apr 2022 at 20:46, Mark Nelson <mnelson@xxxxxxxxxx> wrote:
On 4/12/22 06:27, Radoslaw Zarzynski wrote:
> Hello!
>
> There are 3 ways to deploy crimson.
>
> 1. cephadm. Vanilla images [1] from CI/CD should work though the `cpatch`
> script (and thus `cstart.sh`) is problematic as we build these images with
> different toolchain / options than the classical ones.
> I believe this is the least frequented way at the moment and I would be
> grateful for bug reports.
>
> [1]: eg. https://shaman.ceph.com/builds/ceph/master/9ff904a467c040321318c9ea59e04402030f190d/crimson/301113/
>
> 2. Rook. Some folks already gave it a try and the crimson support went over
> a few rounds of bug fixing. The sole departure from the classical OSD
> is the need to inject the `examples/osd-env-overrride.yaml` [2]
> before deploying
> your OSDs. This won't be necessary when we start to ship crimson images
> without ASan debug facility (production images).
>
> [2]: https://github.com/rook/rook/pull/9647/files#diff-c7f1d89eb9467c9c152c0d03f14a385ee32c889010209ddc869636d3a96a319c
>
> 3. Manual builds. I think this is the most frequented way right now.
I also wanted to mention that if you do a manual build, you can deploy
multi-node test clusters with cbt in a similar way to deploying with
classic. I was planning to do that for quincy on our 10 node mako
cluster but got sidetracked by the allocator regression we are working
on now.
Mark
>
> Regards,
> Radek
>
>
>
> On Tue, Apr 12, 2022 at 3:07 AM Sam Just <sjust@xxxxxxxxxx> wrote:
>> I think current main should have initial crimson support in cephadm.
>> Radek: is that right?
>> -Sam
>>
>> On Mon, Apr 11, 2022 at 12:08 AM Kevin Zhao <kevin.zhao@xxxxxxxxxx> wrote:
>>> Dear community,
>>>
>>> We are working trying to working on Crimson validation and want to test the large cluster deployment and performance for Ceph Crimson.
>>>
>>> Just a quick questions, do we have a traditional method to run the multiple node deployments for Crimson-OSD instead of vstart.sh? Looking through the Cephadm/Ceph-ansible,I don't find one can support crimson-osd.
>>>
>>> Thanks in advance!
>>>
>>> --
>>> Best Regards
>>>
>>> Kevin Zhao
>>>
>>> Tech Lead, LDCG Cloud Infrastructure
>>>
>>> Linaro Vertical Technologies
>>>
>>> IRC(freenode): kevinz
>>>
>>> Slack(kubernetes.slack.com): kevinz
>>>
>>> kevin.zhao@xxxxxxxxxx | Mobile/Direct/Wechat: +86 18818270915
>>>
>>>
>>> _______________________________________________
>>> Dev mailing list -- dev@xxxxxxx
>>> To unsubscribe send an email to dev-leave@xxxxxxx
> _______________________________________________
> Dev mailing list -- dev@xxxxxxx
> To unsubscribe send an email to dev-leave@xxxxxxx
>
_______________________________________________
Dev mailing list -- dev@xxxxxxx
To unsubscribe send an email to dev-leave@xxxxxxx
Best Regards
Kevin Zhao
Tech Lead, LDCG Cloud Infrastructure
Linaro Vertical Technologies
IRC(freenode): kevinz
Slack(kubernetes.slack.com): kevinz
kevin.zhao@xxxxxxxxxx | Mobile/Direct/Wechat: +86 18818270915
_______________________________________________ Dev mailing list -- dev@xxxxxxx To unsubscribe send an email to dev-leave@xxxxxxx
