nova-spicehtml5proxy.service is missing exec reload and task fail

[Kevin Lefevre]

I think there have been some changes in the way services are reloaded:

+ExecReload=/bin/kill -HUP $MAINPID is missing for spicehtml5proxy

systemctl cat nova-spicehtml5proxy.service
# /etc/systemd/system/nova-spicehtml5proxy.service
# Ansible managed

[Unit]
Description=nova openstack service
After=syslog.target
After=network.target

[Service]
Type=simple
User=nova
Group=nova

ExecStart=/openstack/venvs/nova-15.1.4/bin/nova-spicehtml5proxy --log-file=/var/log/nova/nova-spicehtml5proxy.log

# Give a reasonable amount of time for the server to start up/shut down
TimeoutSec=120
Restart=on-failure
RestartSec=2

# This creates a specific slice which all nova services will operate from
# The accounting options give us the ability to see resource usage through
# the `systemd-cgtop` command.
Slice=nova.slice
CPUAccounting=true
BlockIOAccounting=true
MemoryAccounting=false
TasksAccounting=true

[Install]
WantedBy=multi-user.target

[Major Hayden (rackerhacker)]

Which branch/tag did you deploy? I just checked the master, ocata, and newton branches, but I see the ExecReload line in all of those branches.

[Kevin Lefevre (archifleks)]

On Oocata, I think it might be related to my deployment because at some point we switch form spice to novnc.

It seems even though we are not using spice, unit is still there (and maybe tu updated systemd tempate is not copied over because variables is set to novnc) et ansible tries to restart both.

[Jesse Pretorius (jesse-pretorius)]

We currently don't do any clean-up when switching console types. I'm of the opinion that we should just change it to deploy all console types, but set one as a primary. That way switching will be supported, and the option of using more than one console type is available. I'll see if I can find some time to put a patch together to make that happen.