[Gluster-devel] [Release-8] Planning Meeting update
Amar Tumballi
amar at kadalu.io
Wed Jan 22 04:57:11 UTC 2020
Hi All,
We successfully had our Release-8 planning meeting in APAC timezone.
Meeting minutes available here - https://hackmd.io/YB60uRCMQRC90xhNt4r6gA
It was well represented by many component maintainers and many developers
looked enthusiastic to make sure we have a good Release-8 out.
I agreed to setup another Planning meeting for NA/EMEA friendly hours in
coming week, so we can continue with development.
As the above hackmd page would see modification with more content for next
week, I am posting the meeting minutes as part of this email itself.
TL;DR;
-----
Goal
A release of Gluster, which is stable, and provides the functionality to
users, as promised in the documentation. This release also should be a
stepping stone for GlusterX goal
<https://lists.gluster.org/pipermail/gluster-devel/2019-September/056606.html>
References
-
https://lists.gluster.org/pipermail/gluster-devel/2019-November/056709.html
- https://github.com/gluster/glusterfs/milestone/10
Timeline
GA- April 15th, 2020.
RC - 31st March, 2020.
beta - 2nd March, 2020.
Branching - 10th March, 2020.
What is targetted ?
All targetted major issues/enhancements should have the tag of ‘Release-8’
<https://github.com/gluster/glusterfs/milestone/10> in github
<https://github.com/gluster/glusterfs/issues>
Features
The features which need to get into the release must have a Github issue
opened, and have the relevant milestone attached. This helps to track the
issue more closely. It is necessary to have an owner assigned for the
issue, so we are sure whom to ask for progress.
-
Non-Root GeoReplication
- Aravinda Enabling root ssh is not favoured, so better to project
non-root georep (through mount broker)
-
Debugging :
- Error codes Issue #280
<https://github.com/gluster/glusterfs/issues/280> Amar Tumballi
- Instrumentation
-
Logging :
- Structured logging Issue #657
<https://github.com/gluster/glusterfs/issues/657> Yati P
-
Observability :
- More documentation on how to use prometheus or other tools for
Monitoring.
- gstatus <https://github.com/gluster/gstatus> Release 8 compatibility
-
Maintainability:
- Code cleanup (More static functions etc)
- Function scope, extern function reduction etc
- Locked section reduction
-
New Idea Proposals:
- vscan xlator <https://github.com/gluster/glusterfs/issues/265>
- Add more [if you have any]
Bugs / Issues
A good goal is to have less than 200 bugs (<10% high / urgent severity),
and total issues open reduced to less than 200 too.
Current Status:
|Total Bugs
<https://bugzilla.redhat.com/report.cgi?x_axis_field=bug_status&y_axis_field=component&z_axis_field=&no_redirect=1&query_format=report-table&bug_status=__open__&format=table&action=wrap&product=GlusterFS>
|
354 |
|Github issues <https://github.com/gluster/glusterfs/issues> | 434 |
- Bugs to target for Release-8
- https://bugzilla.redhat.com/show_bug.cgi?id=1793042 -
WORM/write-behind.
How can we reduce these?
-
Issues sorted by least active
<https://github.com/gluster/glusterfs/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-asc>
Visit
the top of the issue, and if you can close it with a proper comment and
label WONTFIX, do so.
-
Bugs: If the bug is not critical, and can be resolved with a proper
comment, go ahead!
Notes from the meeting
Introduction by Amar; general setting of expectation from the Gluster 8
release in light of Gluster X discussion from the last meetup. Stability is
the focus. However, keeping documentation and codebase in sync helps our
users and is important to keep in mind.
There is a lot of steps to undertake in order to push the project towards
the intent of Gluster X. Not breaking things; but making features and
capabilities progressive.
Network layer optimization is an improvement that would require intensive
effort to make possible. It should be discussed and planned but not
realistic for release 8.
[Yaniv] Mohit has an experimental patch for performance enhancement (ls -l
etc; anyone has link to the patch?)
Performance improvements are desirable as they have been present for quite
a while in Gluster.
[Aravinda] Path-based GeoReplication - the design is available. It has to
be made to work with sharding. The implementation is not ready yet to make
it a target for release 8; likely a release 9
Multiple set of activities around the topic of Global Threadpool. Amar
hasn’t observed many patches in recent times. Not a release 8
[Yaniv] 3k-4k support has some incremental work available. But a complete
set of work required to get it in release 8 is not yet available. [Amar]
best to think about using some label based tracking mechanism to see how to
test it as entire capability.
RIO has a lot of work at this point and there is not enough full time work
being undertaken on this. Amar has a new approach
<https://docs.google.com/document/d/1KEwVtSNvDhs4qb63gWx2ulCp5GJjge77NGJk4p_Ms4Q/edit?usp=sharing>
that
has received comments from Xavi, and need some more from people with
experience in RIO like Shyam, Kotresh, Susant, so possibly a release 10
candidate.
Code instrumentation and tracing requires a set of changes across the
codebase without a set of designs in place it is not a release 8 target,
but this is critical to get going for future performance improvement work,
would be great to target ‘sooner’ than later.
Yati has been working on structured logging and the given the work the
remaining it is a good release 8 target. Needs review and more
contributions.
- Amar Structured logging is for machines, so don’t expect to have a
Meaningful log.
- Aravinda output will be still readable. The dynamic parts would be
moved into the key/value pair format and are searchable. The messages will
continue to be a prior part in logs.
GFID-split brain resolution has been commented to be a Day1 issue. Unsure
of status. Moving out of release8 milestone
Reviewing the current set of issues at
https://github.com/gluster/glusterfs/issues?q=is%3Aopen+is%3Aissue+milestone%3A"Release+8"
<https://github.com/gluster/glusterfs/issues?q=is%3Aopen+is%3Aissue+milestone%3A%22Release+8%22>
in
order to address what is realistic for the release based on likelihood of
availability
[Yaniv] Issues tagged with release 8 have a higher chance of been addressed
within the time. The topics will be reviewed and appropriately tracked
[Aravinda] Non root geo replication - the documentation would need to be
updated. Migration methods would be handled.
- This is important because many users may not open root access to
remote machine through SSH.
- Non-Root geo-rep is stable, and is ready to be promoted as the default
option now.
[YanivK] Would like to see CentOS8 support, scaled volume counts,
Performance improvement as the major features listed for Release-8.
- The first 2 would be possible to meet.
- The perf improvement challenge would be ‘ambitious’.
[Sankarshan] Why are we seeing only few issues, where are the issues
reported by Users listed in Release-8. To sit with Amar to review the ~400
commits since 7.0
<https://github.com/gluster/glusterfs/compare/v7.0...HEAD> and
review user reported bugs to check for inclusion.
Also need to look at bugs opened by users in the last 4 months to see what
are the few we need to include.
[Sanju] Need help with Cluster-ID design. Once design is finalized, we can
take it up
[Rafi] SHD-Mux - Lot of work has been done on this. Need to take it to
completion. There is a lot of changes on client graph switch. Should we
take this in, if it brings instability?
- Amar Can it be optional?
- Rafi It would be difficult.
- Ravi Can this be done like Sharding?
[Hari] - Working with Shwetha and Sheetal on Build and automation.
https://bugzilla.redhat.com/show_bug.cgi?id=1727727
- How to track this?
- Should I open a github issue? - Yes
[Sunny] Release note for Sunny.
- [sankarshan] to open a release 8 milestone Github issue
<https://github.com/gluster/glusterfs/issues/813> and seek inputs that
should be part of release note - DONE
- to generate graph of Coverity nearer to the release
- the testing framework has shown errors - needs more focus for that to
be stabilized.
[Aravinda]
- gstatus is used still; but not compatible with latest version
- planning to make it compatible with release 8
- monitoring is another aspect which receives requests. A select few
metrics with a Grafana dashboard is something that can be targeted. Plan is
to have a webinar for Gluster monitoring
[sankarshan] Question for later - what about the Gluster8/oVirt part?
----
Please post if you have any questions.
Regards,
Amar
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gluster.org/pipermail/gluster-devel/attachments/20200122/12b02dca/attachment-0001.html>
More information about the Gluster-devel
mailing list