Keystone (and other services) fail to execute parallel
Bug #1597872 reported by
Ken Wronkiewicz
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| kolla-kubernetes |
Fix Released
|
High
|
Ken Wronkiewicz | ||
Bug Description
If you launch multiple containers within the same pod that share the same volume and try to write to the same file on a machine with multiple cores available, they'll tend to try to start up at the same time and open the same file and thus bootstrap fails.
Thus, each pod in a bootstrap job needs to have its own log drive.
(It's dramatically easier to debug, IMHO, if the bootstrap jobs just write to stdout, considering Kubernetes will do centralized log collection and tagging for stdout right out of the box)
| Changed in kolla-kubernetes: | |
| assignee: | nobody → Ken Wronkiewicz (wh-ubuntuone) |
| status: | New → In Progress |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote Fix merged to kolla-kubernetes (master) | #1 |
| Changed in kolla-kubernetes: | |
| importance: | Undecided → High |
| status: | In Progress → Fix Committed |
| milestone: | none → newton |
| Changed in kolla-kubernetes: | |
| status: | Fix Committed → Fix Released |
To post a comment you must log in.
Reviewed: https:/
Committed: https:/
Submitter: Jenkins
Branch: master
commit 5a6c09ff5e90e46
Author: Ken Wronkiewicz <email address hidden>
Date: Thu Jun 16 13:29:12 2016 -0700
Fix keystone bootstrap failure
Keystone was failing to bootstrap because keystone-bootstrap couldn't
get access to /var/log/
pod's /var/log/
another as an emptyDir named kolla-logs-
This only occurs when there are multiple cores, so if you have a one
CPU vagrant, it won't happen.
Change-Id: I09e668f35e8d44
Partially-
Partial-Bug: 1597872