7af9200325
During testing it was found that caching the DC endpoint keystone tokens
consumes an excessive amount of memory on systems with a large number of
subclouds. This is because the raw token data contains data across all
subclouds.
There are two related parts to this commit:
- Re-use the keystone Token across all DC subclouds. This single token
is also cached.
- Since the DC token is a "universal token" (project-scoped), we can
re-use this token for all subclouds instead of getting a new token
for each subcloud.
- The subcloud catalog data also contains information for all
subcloud endpoint URLs (required for the second point below)
- cert-mon utility functions which extract endpoint URLs from token data
are updated to use an existing token rather than requesting a new
token. This reduces system resource usage on large DC systems.
Test Plan:
- PASS: Verify token cache does not significantly impact memory
consumption
- PASS: Verify token re-use in cert-mon utils maintains existing
URL-extraction behaviour
- PASS: Verify cert-mon/keystone resource consumption during cert-mon
daily audits
Failure Path:
- PASS: Verify token expiry is correctly handled
Story: 2008960
Task: 43504
Signed-off-by: Kyle MacLeod <kyle.macleod@windriver.com>
Change-Id: I0327eb50db1cb01a6f57c8675a2436389c0b7c94
|
||
|---|---|---|
| .. | ||
| data | ||
| __init__.py | ||
| test_cert_mon.py | ||
| test_subcloud_audit_queue.py | ||