NOTE: This should no longer be required unless you are running from an older copy of bosh-deployment. See this issue for details.

The BOSH Director (and some BOSH deployments) don't always start up successfully after a reboot (intentional or not). The solution seems to be just a simple monit stop all followed by a monit start all. BOSH Reboot Patch injects that into /etc/rc.local.

When (re-)deploying the BOSH Director, just include the ops file from this repository. Typically, the usage is to simply clone a local copy of the repository, set BOSH_REBOOT_DIR to that directory and then add the ops file like this:

$ bosh create-env ${BOSH_DEPLOYMENT_DIR}/bosh.yml \
    --ops-file=... \
    --ops-file=... \
    --ops-file=... \
    --ops-file=${BOSH_REBOOT_DIR}/operations/add-to-director.yml \
    --ops-file=... \
    --ops-file=... \
    ...etc...

For a fuller example when deploying a director, pleaes see the libvirt-bosh-cpi.

To apply to other deployments, this patch can be added as a runtime config:

bosh update-runtime-config --name reboot-patch \
    ${BOSH_REBOOT_DIR}/manifests/runtime-config.yml

Everything of interest is in the reboot-patch job. That source is here. The templates/pre-start injects the hook into /etc/rc.local while templates/startup.sh is what does the Monit stop/start.