The WebDAV blobstore and Cloud Controller API / Clock / Worker jobs are
unable to start after a VM restart because creation of the run directory
for each process, /var/vcap/data/sys/run/*, was moved to their bosh
pre-start scripts. The jobs are unable to start because
/var/vcap/data/sys/run is mounted on a temporary file-system and the bosh
pre-start script is not executed on VM restart, only deployment. A fix is
in the pipeline for CF-240. To workaround this issue, operators can do a
bosh deploy, which will recognize the failing jobs and properly create the
run directory.

Thanks to Matt Sykes from IBM for reporting this issue.

https://github.com/cloudfoundry/capi-release/issues/25



Nick

--
Nicholas Calugar
Product Manager - Cloud Foundry API
Pivotal Software, Inc.