Dear Tumbleweed users and hackers,

Again I cover two weeks, with the 2nd one being more of troubles than releases. The snapshots covered this week are 0205, 0207, 0208, 0209 (week 6) and 0212 (week 7). I’ll try to explain at the end what was holding us back in week 7.

Noteworthy updates as part of the last snapshots:

  • AppStream-glib update. Most notably: the icons are no longer duplicated across multiple repositories metadata
  • GCC 5.3.1 is now the default compiler
  • Mozilla Firefox 44
  • cmake 3.4.3 – most importantly, ‘modules’ no longer get a soname during linking. Some packages might fail rpmlint now in case the shared library packaging policy was not applied
  • GStreamer 1.6.3 – Many bugfixes
  • Plasma 5.5.4 – you know you want it!

What is brewing in the next snapshot (0217+)

  • Firewalld – available for testers. A future version should bring migration assistance from SUSEFirewall2 to firewalld
  • plymouth gained a new plugin to show simple texts (used to indicate which partition to unlock)

I would like to introduce a new section, mentioning ‘bugs I feel are noteworthy’. The list is incomplete and might not cover the bug YOU had seen, and it shall not be a complete list of new bugs reported.

  • Tumbleweed users using LUKS encrypted devices reported plymouth related issues. Current workaround identified is to disable plymouth at boot time with “plymouth.enable=0” on kernel command line. boo#966255
  • systemd 228 introduced cgroup Task accounting to limit fork bombs and the like. it seems a user logging in on the terminal is more limited than a GUI user, which could result in the system running out of resources to fork more tasks. boo#966878

And, last but not least, as promised, a word about why there have been no new snapshots released since 0209 (released on Feb 12):
openSUSE Tumbleweed relies heavily on openQA and the test results we receive from the automated tests. For not yet clear reasons, the most powerful worker we have (had) on decided to quit its job (it did not mention the wish to find a new career path). This left openQA running with only 2 workers left instead of the usual 10. The remaining work force is largely overloaded and can’t cope with the workload. There were only two options left for us: release without testing or hold back the snapshots. As evidence shows, we opted to hold back the snapshots. Various solutions are currently evaluated to get new worker power on openQA, which includes borrowing machines from other SUSE owned instances.