seed cloud cannot signal 'ready' - we guess at readiness and then race with os-collect-config
Bug #1329528 reported by
Robert Collins
This bug affects 4 people
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| tripleo |
Fix Released
|
High
|
Nicholas Randon | ||
Bug Description
Right now we assume 'can ssh' == 'can start initialisation', but this is incorrect: after ssh starts working, os-collect-config will run and all the services will go through various restarts - without an interlock to ensure it we can end up talking to a service that isn't available: so we need wait_for on *everything* (which seems bogus), or we need some way to emulate 'stack_ready' for the seed.
| Changed in tripleo: | |
| assignee: | nobody → Nicholas Randon (nicholas-randon) |
| status: | Triaged → In Progress |
Revision history for this message
| OpenStack Infra (hudson-openstack) wrote Fix merged to tripleo-incubator (master) | #1 |
| Changed in tripleo: | |
| status: | In Progress → Fix Released |
To post a comment you must log in.
Reviewed: https:/
Committed: https:/
Submitter: Jenkins
Branch: master
commit 587515f153b0d43
Author: Nicholas Randon <email address hidden>
Date: Thu May 29 17:23:20 2014 +0100
Wait for os-collect-config to complete on the seed
Currently in devtest we wait for a node to become pingable and then
start configuring the seed details. However the node may not have
completed its configuration as os-collect-config may still be running
on the seed node, configuring low level services. This means we can
start configuring services too early and fail to correctly configure
the seed.
As the seed image carries local metadata and it is therefore not
under Heat control, we need to listen for and send a completion
signal. This allows us to detect when the seed has completed
configuration and the jumphost can safely continue.
Requires for CI: Ib52efa0f3f7c26
Change-Id: Ibfd20bf632cdf3
Closes-Bug: #1329528