VIM patch strategy unlock timeout sometimes occurs
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
StarlingX |
Fix Released
|
Low
|
Al Bailey |
Bug Description
Brief Description
-----------------
In a large environment (many computes) if many hosts are patched in parallel, the unlock commands may timeout due to the operation taking longer than 45 seconds to generate the hieradata.
Severity
--------
Minor
Steps to Reproduce
------------------
Try to unlock 10 hosts in parallel as part of a VIM patch strategy.
Expected Behavior
------------------
It should work
Actual Behavior
----------------
VIM sometimes reports a failure and you need to run the patch strategy again to ensure any remaining hosts are patched.
Reproducibility
---------------
This scenario and environment are not common. However it can be reproduced with 10 hosts in parallel.
System Configuration
-------
STD Duplex with many computes (and can include storage too)
Branch/Pull Time/Commit
-------
This has always been like this. Lets say June 2022.
Last Pass
---------
It sometimes passes and sometimes fails, all due to what the load is like during hiera generation.
Timestamp/Logs
--------------
Here is a 50 second sysinv-api REST operation to issue the 'unlock'
sysinv 2022-08-18 14:12:30.341 117610 INFO sysinv.
Test Activity
-------------
Feature Testing
Workaround
----------
Run the strategy again
I am going to just increase the 45 second sysinv API timeout in the VIM from 45 seconds to 60 seconds.
Changed in starlingx: | |
assignee: | nobody → Al Bailey (albailey1974) |
importance: | Undecided → Low |
tags: | added: stx.8.0 stx.nfv |
Fix proposed to branch: master /review. opendev. org/c/starlingx /nfv/+/ 853677
Review: https:/