Fix #4913
closed
don't try and fetch non-existent packages
Added by Greg Farnum about 11 years ago.
Updated about 10 years ago.
Description
Sage has noted several failures in the nightlies that stem from trying to fetch packages out of the notcmalloc that aren't there yet, and I observed this over the weekend on an FS nightly. We should block starting the test until they do exist.
- Priority changed from Normal to High
- Tracker changed from Bug to Fix
- Target version set to sprint6
- Assignee set to Anonymous
- Translation missing: en.field_story_points set to 5.0
What does the wait_for_package parameter to install() do?
What does the wait_for_package clause for install() do?
Update: It looks like the wait_for_package clause may have already been implemented to handle this.
The ability to wait until all packages are available has been implemented. At first, I thought that some of the suites may need to be changed to use this feature. However, since the suites have run without this being an issue, it may be the case that the current set of tests do not need to use this.
- Status changed from New to Resolved
- Translation missing: en.field_story_points changed from 5.0 to 2.0
After reviewing this with people and looking at the code, I've come to the following conclusions:
1. There used to be a problem with notcmalloc packages taking much longer to fetch. This does not appear to be the case now.
2. A wait_for_package yaml clause can be used to cause teuthology to wait for a package to become available. The method to change a test to do the wait is implemented.
3. No yaml file in ceph-qa-suite uses wait_for_package. It appears that none of the current tests are hampered by this problem.
Based on this, I am answering this problem as being resolved.
Also available in: Atom
PDF