Check that worker-multiplier is set
Bug #1994977 reported by
Giuseppe Petralia
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Juju Lint |
Fix Released
|
Wishlist
|
Gabriel Cocenza | ||
Bug Description
Most of openstack API units have a worker multiplier. If this is unset it will default to 4 workers for each API units. Most of clouds have at 0.25.
Default 4 workers is too less for a production cloud and we saw random http 502 when this was unset.
Juju lint should raise a warning/critical if this is unset especially for keystone, neutron-api, nova-c-c, cinder, octavia, placement, glance
Also nova-compute has it and we usually set to 0.1
Related branches
~gabrielcocenza/juju-lint:bug/1994977
- 🤖 prod-jenkaas-bootstack: Approve (continuous-integration)
- Erhan Sunar (community): Approve
- Mert Kirpici (community): Approve
-
Diff: 139 lines (+75/-1)4 files modifiedcontrib/includes/networking/ovn.yaml (+5/-0)
contrib/includes/networking/ovs.yaml (+5/-0)
contrib/includes/openstack.yaml (+39/-1)
tests/unit/test_jujulint.py (+26/-0)
| description: | updated |
Revision history for this message
| Nobuto Murata (nobuto) wrote | #1 |
Revision history for this message
| Giuseppe Petralia (peppepetra) wrote | #2 |
| Changed in juju-lint: | |
| importance: | Undecided → Wishlist |
| Changed in juju-lint: | |
| status: | New → Fix Committed |
| assignee: | nobody → Gabriel Angelo Sgarbi Cocenza (gabrielcocenza) |
| milestone: | none → 1.0.5 |
| Changed in juju-lint: | |
| status: | Fix Committed → Fix Released |
To post a comment you must log in.
Please note that 4 workers are not ideal but a fixed number as multiplier was also problematic.
We need to pay attention to resource consumption of higher number of workers and also mysql connections.
For example with some AMD CPUs, 0.25 can be 64 workers per one service (64 cores per socket * 2 threads per core * 2 sockets) and one unit can run multiple services.
References:
https:/
https:/