action #12982
closed
What are typed by type_string on ipmi physical machine is not correct.
Added by xlai almost 8 years ago.
Updated about 7 years ago.
Category:
Bugs in existing tests
Description
Job link: https://openqa.suse.de/tests/overview?distri=sle&version=12-SP2&build=2016&groupid=46
Testsuites:
*gi-guest_sles12sp2-on-host_sles12sp2-kvm:
Fail stage: host installation
Fail reason: what are typed out by type_string is not complete
At the last step of installation , 'install and reboot', command 'save_y2logs /tmp/y2logs.tar.bz2 ' is typed to 'save_y2lgs /tmp/y2logs.tar.bz2' which results to 'command y2lgs not found' and exit
Also failed for the similar reason in following build tests.
The type speed is already the max interval, but still type wrong.
The issue has been happening in all recent builds.
xlai, can you further investigate or delegate to someone who could check if IPMI is broken, outdated firmware or other ideas?
I used to open a issue for os-autoinst about this problem https://github.com/os-autoinst/os-autoinst/issues/555, but kulow closed it. And he also claims that the keyboard driver for ipmi is buggy. I think this is the root cause. So I do not think it can be fixed unless supermicro ipmi fixes the buggy problem.
I know that the firmware of ipmi is outdated in nuremburg, however our beijing machines are the latest, it also has ipmi issue that the serial console connection usually breaks. SO I do not think upgrading the firmware can fix the problem.
I have also talked about it with kulow and my manager calen. There will be some discussion at manager level.
What's the usage of detecting it? Retype again when detecting typing error? Even with this, there must be maximum try times set, but still can not ensure it is correct before final retyping.
Yes, exactly what you wrote. Something like
- try typing
- if wrong, try again up to 7 times
- if still fails for 7 times, abort
The challenging part would be to detect errors in typing. I am thinking of calling the last command typed with script_output and checking against expected. With fc -n -l -1 one could read the last command executed. This could be compared against what was requested to be typed. If it differs, try again.
- Blocked by action #12926: Can not get ipmi serial output correctly. added
- Status changed from New to Closed
My position was correct, the problem is not the testcase but something in the IPMI backend.
Closing as the related one is more relevant.
- Status changed from Closed to New
- Category set to Bugs in existing tests
- Status changed from New to Closed
This is because the ipmi keyboard driver does not work stablely under high pressure. We have tried to decrease the pressure as much as possible in our tests, but it is still not completely avoidable.
So close it.
This is an autogenerated message for openQA integration by the openqa_review script:
This bug is still referenced in a failing openQA test: gi-guest_sles12sp1-on-host_sles12sp3-kvm
http://openqa.suse.de/tests/790299
This is an autogenerated message for openQA integration by the openqa_review script:
This bug is still referenced in a failing openQA test: gi-guest_sles12sp1-on-host_sles12sp3-kvm
http://openqa.suse.de/tests/790299
This is an autogenerated message for openQA integration by the openqa_review script:
This bug is still referenced in a failing openQA test: virt-pvusb-sles12sp1-pv-on-sles12sp3-xen
https://openqa.suse.de/tests/873791
Also available in: Atom
PDF