This task is primarily meant to capture the upstream work, which consists of development activities and patch review that happen outside the RHEL process.

July 9th, 2024

• Reworked my miscellaneous multipath fixes patchset based on Martin’s review
• Working on a fairly substantial change to the multipathd path checking code based on a discussion I had with Martin. The goal is to both make multipathd get a more coherent idea of its paths state, and spread out the path checking workload of the different multipath devices. Unfortunately, there are lots of corner cases in this part of the code. Doing lots of cleanup and fixing minor bugs as I go. Currently it's 18 patches long, and it’s almost done, but it will need some targeted testing before I’m sure enough to post it.
• Need to get an annoying multipath related dracut change reverted. Dracut development is kinda weird.

July 23rd, 2024

• Reviewed Martin’s 49 patch upstream multipath set.
  • Required rebasing Ben's own patchset
• Finished working on my path checking patchset, spent a while testing the corner cases, and reworking it based on Martin’s reviews. He’s now on vacation, but I’m pretty sure that he will accept v3 of the last three patches.
• Fixed two bugs in mpathpersist that make preempt and abort commands fail when they shouldn’t and register commands succeed when they shouldn’t.

Aug 6th, 2024

• RHEL-28068 & RHEL-44569: Fixed and in errata
• RHEL-49747 fix posted upstream
• 3 OSCI tests repeatedly failed apparently due to udev events getting handled reeaallly slow on the CI machines. Ended up rewriting the tests.
• Also made it possible to get some more status information from multipathd so it will be easier to detect this.
• Noticed another issue where multipathd isn’t correctly retrying failed removes of in-use devices. Working on a fix.

Sept 3rd, 2024

• Reviewed yet another 40+ upstream multipath patchset. It was mostly a straightforward cleanup.
• Did the other part of my checker rewrite, that my previous checker work made possible. Multipathd now starts the checkers on all the paths that are due for a check, then handles any uninitialized paths needing work, then comes back and gets the results of all the checkers, and finally updates the changed multipath devices once based on all the path changes. Posted upstream and waiting for reviews.

Sept 17th, 2024

• More checker work. In my discussions with Martin Wilck about the last version of my patchset, I realised that I could do one more optimization. All the checker work had brought the code close to the state where we could drop a lock that it was holding while it briefly waited for the asynchronous checkers. DId the work to make multipathd handle cases where multipath devices or paths get changed while we are in the middle of checking a path. Submitted the patches upstream and waiting for reviews