Deployment of MOS 9.1 with StackLight - InfluxDB/Grafana Fuel Plugin is failed
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
StackLight |
Fix Released
|
High
|
Simon Pasquier | ||
0.10 |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
Summary:
systemd package was recently added to trusty-updates repository. That's why post-install script from influxdb package tries to use systemclt to enable influxdb service. That leads to deployment error.
Env description:
- compute
- cinder+controller
- elasticsearch_
Steps to reproduce:
1. Get Fuel 9.1
2. Configure
- compute
- cinder+controller
- elasticsearch_
3. Deploy environment
4. Deployment ends up with the following error: http://
This issue occurs because post-install script tries to use systemcl to enable influxdb service. Here is the parts of the post-install script which lead to the issue:
...
function install_systemd {
cp -f $SCRIPT_
systemctl enable influxdb
}
...
...
elif [[ -f /etc/debian_version ]]; then
# Debian/Ubuntu logic
which systemctl &>/dev/null
if [[ $? -eq 0 ]]; then
install_
else
# Assuming sysv
install_init
install_
fi
...
P.S. As I've figured out from here https:/
pre_build_hook just downloads influxdb deb package from https:/
So we need to patch that package and place it somewhere else.
[1] https:/
Workaround #1:
=============
Add systemd-shim to "additional packages" in fuel interface before deploying the environment.
Workaround #2:
=============
1. Provision the environment.
2. Install systemd-shim on the influxdb-grafana nodes.
3. Deploy the environment.
Changed in lma-toolchain: | |
status: | New → Confirmed |
tags: | added: customer-found |
tags: | added: blocker-for-qa |
Changed in lma-toolchain: | |
assignee: | nobody → Swann Croiset (swann-w) |
assignee: | Swann Croiset (swann-w) → guillaume thouvenin (guillaume-thouvenin) |
Changed in lma-toolchain: | |
importance: | Undecided → Critical |
milestone: | none → 1.0.0 |
description: | updated |
description: | updated |
I believe I am hitting the same issue with Stacklight in MOS 8 w/MU3. Ping me directly via Slack if you need access to lab where deployment is failing. See paste to match above:
http:// paste.openstack .org/show/ 593479/