action #43877
closed
[ha] test fails in consoletest_setup - Jobs on ppc64 end up without ip on MM tests
Added by ldevulder over 5 years ago.
Updated over 5 years ago.
Description
Observation¶
openQA test in scenario sle-15-SP1-Installer-DVD-ppc64le-ha_beta_node02@ppc64le fails in
consoletest_setup
Reproducible¶
Fails since (at least) Build 58.1
Expected result¶
Last good: (unknown) (or more recent)
Further details¶
Always latest result in this scenario: latest
The test fails because there is no network on the VM. IP address should be set by the DHCP server running on the support-server (this test in that case: https://openqa.suse.de/tests/2263783).
But sporadically the IP address is not set (in fact I more often have no IP than one...). This happens on all MM ppc64le workers (malbec and QA-Power8-5-kvm).
Looks like ppc64le specific, as I have no problem with MM jobs on x86_64 and aarch64.
- Project changed from openQA Tests to openQA Project
- Subject changed from test fails in consoletest_setup to [u] test fails in consoletest_setup - Jobs on ppc64 end up without ip on MM tests
- Category deleted (
Bugs in existing tests)
- Assignee deleted (
szarate)
@szarate should we really keep this ticket on the backlog of QSF-u? I see this highly unrelated
@okurz I want to at least give it a look :) Whether we do it or not, it's another story
sure, propose a duration for time-boxed work, I'd say
- Subject changed from [u] test fails in consoletest_setup - Jobs on ppc64 end up without ip on MM tests to [ha] test fails in consoletest_setup - Jobs on ppc64 end up without ip on MM tests
- Category set to Support
- Assignee set to ldevulder
@ldevulder, seems this is something "ha" specific?
okurz wrote:
@ldevulder, seems this is something "ha" specific?
MM specific more (network issue before HA stack), but as on ppc64le only HA uses MM we can say HA specific :)
As I don't have any IP address it may be an issue with the DHCP server in the supportserver or a configuration issue on the worker? But as it works sometimes I think it's more an issue on the worker like we have/had sometimes also on other arch (like aarch64 for example during SLE15 BV).
- % Done changed from 0 to 70
After investigation I found that GRE tunnel was not started on malbec, so jobs in the same group but running on different workers were not able to connect to each others.
A reboot of malbec is planned to resolve this.
- Status changed from New to In Progress
- % Done changed from 70 to 80
Server malbec has been rebooted, ppc64le MM tests in progress.
- Status changed from In Progress to Resolved
All is OK now, so ticket set to Resolved.
Also available in: Atom
PDF