Ubuntu Server team update - 22 July 2019

Hi everyone, below you will find the updates of the Ubuntu Server team members from the last week. If you are interested in discussing a topic please start a thread in the Server area of this Discourse site.

1 Like

Short week due to vacation Thursday/Friday last week, and some of last week was spent catching up from partner visits the previous week. :sun_with_face: Summer time! :sun_with_face:



  • Initial review of per-series apt keys, follow-up review of Chad’s changes to come this week


  • cloud-init/curtin/streams bug triage on Monday, Tuesday & Wednesday
    • It was my week for the rotating shift, and vacation meant that someone needed to pick up a spare day
  • Reviewed a jenkins-job-builder Jinja2 change from Phil on the Certified Public Cloud team
    • I contributed the original Jinja2 templating work upstream, which is why Phil sought my review
    • In the course of this, we discovered that I had misnamed something in my original work, which is annoying; naming is hard!
  • Validated my SRU fix for a tmux memory leak which was released in to bionic (the only affected release) late last week
  • Discovered this great series of posts where Darius Kazemi summarises one early RFC per day
    • (I was glad to discover the authors of RFC-48 also screwed up the naming of something, so I’m just joining a long tradition of screw-ups :grin:)


  1. QEMU HW mitigations support (ARCH_CAPABILITIES)
    LP: #1828495

Is already in disco-proposed (1:3.1+dfsg-2ubuntu3.3) and I’m verifying Disco and Bionic -proposed packages. Will provide feedback today. We’re still working on the libvirt support for those.


  1. Pacemaker and Corosync
    Eoan Migrations - I’m cleaning up all bugs/issues blocking Ubuntu HA related packages (corosync, pacemaker, crmsh, etc) that are either in [main] pocket or in [universe] but important to HA (like crmsh).

  2. For reverse dependencies in corosync, I’ve cleaned up corosync itself, dlm, gfs2-utils, ocfs2-tools. I still have to fix corosync, crmsh, pacemaker and sbd regressions.


  • Corosync merges were blocked by regressions. Corosync depends on libknet1 and it wasn’t in [main] pocket. We’ve asked for a MIR and it has been accepted and finished.

Corosync and Pacemaker

  • Pacemaker merges were blocked by a alleged regression for armhf architecture. I have tracked this down to be an issue with armhf lxd non privileged containers and created a verification in order to run or skip autopkgtests in such conditions.

** OTHER **


Release all the things week! Triage all the things week!




Short week due to vacation.


  • php 7.2 to 7.3 transition
    • Improved error handling for build scripts
    • Reviewed remaining build errors, most of which are due to requiring intermediate dependencies, so for the PPA will need to structure the order they’re uploaded.
    • Discussed background of past transitions with Nish, and got his advice on dealing with some of the common built issues.
  • LP: #1790657 SRU for librabbitmq, “server parameters unusable”
    • Merge proposal was reviewed and accepted, and uploaded the package for xenial, bionic, & cosmic
  • LP: #1804542 SRU for bind9, “Multiple intermittent socket failures during name resolution”
    • Prepared ppa with a backport I made of the upstream commit that purportedly fixes the issue.
    • Requested reporters help identifying how to reproduce the issue more definitively, and to test the PPA to assure it solves the problem.
  • LP: #1797836 SRU for logwatch, “logwatch reports false kernel error warning”
    • Investigated how to synthetically insert the error in an lxd container, to facilitate validation of the fix. What makes it a bit tricky is that this is a kernel log, but within LXD the kernel logs are provided and owned by the host, thus inaccessible, so will need to experiment a bit further. On the plus side, once there is a reliable way to handle this, it should be applicable to most all other logwatch bugs and may be usable for some new autopkgtests.


  • Ubuntu Maintainers’ Handbook
    • Started drafting bug triage docs in Ubuntu Maintainers Handbook, to replace its stubbed in page. The team also maintains some bug triage documentation in wiki, which potentially could also be merged into the handbook.


  • numad bionic and disco SRUs for LP: #1832915 - Approved
    • Was not able to reproduce the issue for testing, but had some review comments which were addressed.
  • dpdk fix for LP: #1836365 - Approved
    • Verified debian delta are good, and that SRU policy for DPDK is being followed.
  • apache2 bionic and cosmic SRU for LP: #1836329 - Approved
    • Reproduced the CPU spin 100% bug for bionic. Required setting up port forwarding on my home router into an LXD container running the problematic code. Verified patch fixed the reported issue.


team procedures

  • Improved the Server Team daily triage email for curtin, cloud-init and simplestreams to include old New bugs.


  • review an MP for a new test case for the the “reuse half a raid” scenario.

performance metric

  • boot speed measurements: measure the Core16 beta images on devices


  • testing of the new “reuse existing partitions feature”
  • the testing resulted in me filing bugs against the subiquity project


  • New Ansible configuration for ppc64 ISO testing node
  • New configuration to allow the Jenkins worker nodes to reliably send local and remote email