LWG Minutes 2019-08-22
Cray: Cory Spitz
ORNL: James Simmons, Dustin Leverman
HPE: Christopher Voltz
SuperMicro: Abe Asraoui
CEA: Jacques-Charles Lafoucriere
Intel: Ralph Castain
Linaro: Joshua Terry, Elsie Wahlig
Whamcloud: Joe Gmitter, Peter Jones, James Nunez, Nathaniel Clark
New Actions Captured:
Existing Open Actions:
Actions Recently Closed:
2.13.0 Release Update
- Things continue to progress with landings. Oleg has seen an issue with the latest batch of master-next. Oleg is narrowing it down and testing a fix. There is about half a dozen of patches that may have be the culprit.
- A gentle reminder that the documentation tasks needed for release. Doc tickets can be tracked with this filter: https://jira.whamcloud.com/browse/LUDOC-455?filter=15718#.
- Anything else to report out on?
- Cory: Just a heads up that we found a blocker issue with SEL and will be pushing a fix under LU-12681.
Upstream Lustre Client
- Neil has been moving forward, almost at 2.12 LTS. James will be going through all of his ports and reviewing.
- Ken is not on the call today, but did send in some comments.
- For lustre.org there have been subsequent security-related updates required to the main site that have now been completed without the need for downtime.
- For wiki.lustre.org i'm in the process of getting us moved to the most recent LTS release (1.31.3) and expect to have that completed over the next week. This will require a brief period of downtime (in the past usually ~5mins) and will perform this during a low-traffic period.
- The discussion continues from last meeting regarding what to do when a patch fixes an issue in RHEL/CentOS and would apply to other distros that the developer does not have access to validate.
- Peter: If nobody is willing to stand up and take on the responsibility to regression test other distros, then it is not realistic to expect a developer to have to take on that burden without access to those test environments. We should not hold back a patch from landing that fixes something in RHEL/CentOS just because it is not worked out in other distros.
- No objections raised to this approach.
- Peter: We have added in ARM clients to the CI system and have testing running. There are some tests that had to be turned off because they were always failing. It wasn't clear if the test was making wrong assumptions for ARM or if there is a bug. James, do you have any input into these tickets?
- LU-11729: James is looking at it. Needs more than just Andreas’ fix. James believes this is a bug in the kernel, but can’t say for sure until digging deeper.
- LU-10073: James pushed a patch for Amir to try and it has mixed results. Amir has a theory that it could be something between the host and guest VMs.
- LU-11596: James has seen this one and hit in the upstream kernel as well.
- LU-11671: James has not looked at this one. We have seen it on ARM and it is turned off.
- LU-11667: Andreas thinks it might be a test script issue, but nobody tested this theory yet. James has not tried it either.
Next meeting will be on 2019-09-04 at 11:00am Pacific