read_timeout not supported by pymysql causes deployment failures
Bug #1526938 reported by
Alex Schultz
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Released
|
High
|
Alex Schultz |
Bug Description
Currently we have a read_timeout=## added to our connection string settings[0] within fuel library.
Unfortunately pymysql does not support this option[1] which results in a stack trace & failure if the configuration has the read_timeout defined and pymysql is used.
Currently the UCA packages use pymsql instead of python-mysqldb as used by the MOS packages. We need to only provide the read_timeouts if supported by the python package being used.
[0] https:/
[1] https:/
Changed in fuel: | |
status: | New → Confirmed |
Changed in fuel: | |
assignee: | Alex Schultz (alex-schultz) → Andrey Shestakov (ashestakov) |
Changed in fuel: | |
importance: | Medium → High |
Changed in fuel: | |
assignee: | Andrey Shestakov (ashestakov) → Alex Schultz (alex-schultz) |
tags: | added: team-enhancements |
Changed in fuel: | |
assignee: | Alex Schultz (alex-schultz) → Matthew Mosesohn (raytrac3r) |
Changed in fuel: | |
assignee: | Matthew Mosesohn (raytrac3r) → Ivan Berezovskiy (iberezovskiy) |
Changed in fuel: | |
assignee: | Ivan Berezovskiy (iberezovskiy) → Alex Schultz (alex-schultz) |
tags: | added: on-verification |
tags: | removed: on-verification |
Changed in fuel: | |
status: | Fix Committed → Fix Released |
To post a comment you must log in.
@Alex, keep in mind that this option is not explicitly supported in oslo.db, but it is in sqlalchemy. I think we should file an upstream bug to pymysql for the regression (since it's a rewrite of python-mysqldb). Without the read_timeout variable, it will take I belive around 120s for connections to timeout, which is less than ideal.