rgw team retrospective

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



the rgw team had a retrospective meeting this morning about our
upstream processes, and came up with a lot of good ideas for
improvement. thanks to everyone that joined! i'm also interested to
hear how other teams are handling similar issues

## topics of discussion

1. subject matter experts
2. tracker issues
3. pr review
4. rgw teuthology suite
5. pr testing


## subject matter experts

an idea that came up repeatedly was that of a designated 'subject
matter expert' for each part of rgw, who would take a bigger role in
reviewing and delegating the bugs and pull requests

the team has a lot of great engineers with deep, specialized knowledge
of rgw, and i'd love to find more ways to spread that knowledge around
through collaboration and mentorship

i'll be creating a document, probably as a pull request to the ceph
repo, that breaks down these areas and names the maintainers


## tracker issues

the rgw project has a lot of open tracker issues. we have a ~30 minute
bug scrub every thursday, but many bugs remain unassigned and we end
up scanning the same recent issues every week without making much
progress

* assign every bug! default to the relevant expert, who can delegate
as necessary

* make better use of tracker priorities to focus on a) severe user
issues and b) teuthology failures

* occasionally do 'deep scrubs' to cover the really old stuff, and
either close the bugs or assign to an expert for further review

* developers can work with their manager to help prioritize bug
fixing, and to make sure nobody is being overwhelmed with too many
assigned bugs


## pr review

* pr reviews can be assigned to their subject matter experts,
especially for old/stale prs that need attention

* for pull requests authored by experts, try to get another team
member involved early on the in process, so they can participate in
design discussions and have a better understanding of the
project/feature

* very large pull requests can be broken into smaller pieces to make
the review more accessible. it also helps to put some extra effort
into documentation, code comments, commit messages, and pr description

* face-to-face code reviews are a good way to reduce communication
overhead, and can help focus review on the interesting parts

* i encourage everyone to participate in code reviews on github, even
if they're not very familiar with the features. following pr
discussions is a great way to absorb knowledge, and a fresh
perspective can help to catch big-picture issues that the author
missed


## rgw teuthology suite

we've been struggling with a lot of test failures since the pacific
release. a lot of them are due to the general entropy of changing test
environments, and others are inevitable from our larger refactoring
projects

the failures make it difficult to review teuthology results, because
there are so many 'known issues' to keep track of. and sometimes those
mask other failures, so don't prevent us from merging more breaking
changes

track teuthology failures under 'urgent' priority and team up on
fixing them until the suite is green again!


## pr testing

a lot of reviewed prs are tagged needs-qa but go stale without
testing. subject matter experts can identify these old prs that should
be prioritized

let's start cycling through the responsibility of testing pr batches
every week or two? that person can get a handle on the known failures,
and make sure there are tracker issues for everything

if we get the suite green again, that makes it so much easier to get
more people involved in testing

_______________________________________________
Dev mailing list -- dev@xxxxxxx
To unsubscribe send an email to dev-leave@xxxxxxx



[Index of Archives]     [CEPH Users]     [Ceph Devel]     [Ceph Large]     [Information on CEPH]     [Linux BTRFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux