Openstack commands hold prompt > 30 seconds
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
StarlingX |
Won't Fix
|
Medium
|
Tao Liu |
Bug Description
Brief Description
-----------------
During sanity execution, several openstack commands are used (openstack server, openstack image, openstack volume, etc.). Randomly, some of the commands take several seconds to complete and return the shell. This breaks automated sanity execution when it exceed 30 seconds.
Severity
--------
Provide the severity of the defect.
Major: On automated testing, this breaks executions. An automated re-execution or manual is a good workaround to this issue.
Steps to Reproduce
------------------
After an environment is fully provisioned. This can be reproduced by using any openstack command, they usually takes 1-2 seconds to return the prompt. From time to time, it can take >30 seconds to complete and return prompt, which breaks automation. This happens with different configurations, and different openstack options. This is not tied to the actual action, like creating a VM which can take several minutes, the lack of the prompt is what breaks the automation.
Expected Behavior
------------------
After executing an openstack command, prompt should be available in 1-2 seconds.
Actual Behavior
----------------
Execute an openstack command, prompt returns after >30seconds.
Reproducibility
---------------
Intermittent. It happens randomly, but the frequency has increased in the last couple of weeks.
System Configuration
-------
Baremetal and virtual environments. Simplex, Duplex, Standard, Standard-External.
Branch/Pull Time/Commit
-------
Using latest ISO from 20190723T141142Z. This issue is recurrent since July/04. Across all configurations and environments, we execute around 400 openstack test cases daily. This issue breaks on average 10% of them. As mentioned before, we re-execute the test cases (manually or automated) and the test case passes.
Last Pass
---------
As a reference, I checked on our old logs, this issue is recurrent since first week of July, but it seems to be increasing. Before that date, this happened just a handful of times.
Timestamp/Logs
--------------
Here are some examples of the commands that are failing:
http://
I'm not sure if a collect will actually help to debug, let me know and I'll provide it. With the next build, I'd use the healthiest configuration to ran some commands several times and capture the time that it takes to return the prompt.
Test Activity
-------------
Sanity.
tags: | added: stx.sanity |
Changed in starlingx: | |
status: | Triaged → In Progress |
@Christopher, can you provide the collect logs from the standard system when the openstack show cmd takes more than 30 seconds? Please note the timestamp for the cmd in the notes.