Closed sykesm closed 8 years ago
We have created an issue in Pivotal Tracker to manage this:
https://www.pivotaltracker.com/story/show/126688091
The labels on this github issue will be updated when the story is started.
@sykesm im assuming you are on softlayer. we have recently fixed that problem in the agent: https://github.com/cloudfoundry/bosh-agent/commit/b74fb972dc671bcabc4cabe2f04d2e08af74a08c. i dont believe it's made into any stemcell version yet.
@cppforlife yes, on SoftLayer. We did see the pending change to the agent but, as you said, it's not in any available stemcell.
That will help correct this particular problem but we hit several others related to pre-start when SoftLayer rolled VMs in a pod to apply security fixes. One example is in cloudfoundry/capi-release#25.
@sykesm can I close out this issue, as you expect a fix in the upcoming stemcell?
Sure.
The consul agent's pre-start script appends
127.0.0.1
to/etc/resolvconf/resolv.conf.d/head
and the bosh agent in stemcell 3262.2 writes the dns server entries to the same file.When a system running the consul agent is rebooted, the pre-start script does not run but the agent initialization scripts do so the modifications made to /etc/resolvconf/resolv.conf.d/head are lost and
127.0.0.1
is no longer in/etc/resolv.conf
.The net result is that any job that relies on the consul for DNS resolution fails after a reboot.