kernelci.org update - 2020-11-06 #minutes

kernelci.org update - 2020-11-06 #minutes

[Guillaume Tucker]

Summary of changes going into production
========================================

* LTP IPC tests run on a range of x86, arm64 and arm platforms
* New device imx6qp-wandboard-revd1 in lab-pengutronix
* Disable kselftest in lab-collabora due to long job timeout (2h)
* Recreate indexes without "background" option for Mongo DB 4.2


Technical Steering Committee minutes - 2020-11-03
=================================================

* Timeslot is tied to UTC.  Is this slot still working for everyone?
  -> will move back to 5pm UK time so it follows DST

* KCIDB

  KCIDB revision objects are not really revisions, they're occurences of
  revisions (multiple objects can represent the same commit in multiple repos,
  or the same patchset in multiple maillists). However, they are identified
  with revision IDs.

  This makes it impossible to aggregate gradually-, or repeatedly-reported
  "revision occurrences" without losing the information on multiple sources.

  See an example of such a revision:
  https://staging.kernelci.org:3000/d/revision/revision?orgId=1&var-dataset=kernelci04&var-id=3650b228f83adda7e5ee532e2b90429c03f7b9ec

  So, we need to rename "revisions" to something else, and change their IDs
  again. The proposed changes are included into a draft PR and rename
  "revisions" to "sources", change their IDs to be fully up to origins (again),
  and add the "revision_id" field to them with the syntax of the previous "id"
  We could have a problem with this approach if e.g. the origin tests the same
  source commit multiple times and reports it with different IDs Needs more
  thinking and discussion

  -> to be discussed in further detail over mainling list

* GPL compliance

  Should we add a custom footer on the nginx index to provide information about
  where to get the kernel source and the license text on storage.kernelci.org?
  https://groups.io/g/kernelci/topic/license_compliance_with/77943380
  -> yes

* kernelci.org disk space

  Running out of disk space on the root partition on the kernelci.org server…
  This can be resolved by moving the Mongo DB data to the data partition.  It
  would seem necessary at some point to archive old data in Mongo DB, to move
  some documents to separate collections or dumped to a tarball.  It would
  avoid the issue of running out of space and keep the data size under control.
  -> agreed

* kernelci.org results into KCIDB

  We should be able to retrieve the old data if and when needed, also it should
  get pushed to KCIDB.

  -> sounds like a plan Submitting data to KCIDB with streaming JSON: Kevin to
     start looking into this in the coming weeks


Technical Steering Committee minutes - 2020-10-27
=================================================

Very short meeting in the middle of OSS/ELC Europe 2020.

* login test cases:
  * Fixed issue with auto-login-action rename in LAVA
  * Workaround for kernel errors detected as “login”: take overall job status
    into account


Advisory Board minutes - 2020-10-28
===================================

* Discussing Windows quality tools with Microsoft to compare with Linux kernel
* Start preparing 2021: budget, memberships and general project updates


Advisory Board minutes - 2020-11-04
===================================

* Discussing OSS/ELC Europe, blog post getting published
* Follow-up from the previous Microsoft discussion about kernel metrics


Best wishes,
Guillaume