Cloud provider authentication failures resulting in memory/goroutine leak
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Fix Released
|
High
|
Simon Richardson |
Bug Description
The logs from a Juju 2.5.4 HA controller used for JAAS repeatedly logs:
2019-06-18 01:43:52 ERROR juju.worker.
our systems and retry., accessNotConfigured
2019-06-18 01:43:52 ERROR juju.worker.
our systems and retry., accessNotConfigured
2019-06-18 01:43:52 ERROR juju.worker.
This is presumably due to the GCE account and/or project being removed outside of JAAS. While Juju should detect this case and disable the model, the bigger issue is that memory consumption appears to continually grow linearly on the controller where these messages are logged. When the controller stops logging these messages the memory consumption appears to drop and it will then start to grow on another controller.
tags: | added: canonical-is |
Changed in juju: | |
status: | New → Triaged |
importance: | Undecided → High |
assignee: | nobody → John A Meinel (jameinel) |
Changed in juju: | |
milestone: | 2.6.4 → 2.6.5 |
status: | Fix Released → Fix Committed |
Changed in juju: | |
status: | Fix Committed → Fix Released |
Juju heap profile and goroutines are available at https:/ /private- fileshare. canonical. com/~jsing/ lp1833155/ - it is worth noting that there is minimal heap in use, however a significant amount of stack appears to be allocated.