QA (public) » openQA Project (public) » openQA Tests (public) » qe-yam

• tried MULTIPATH=1 to click the multipath page and get the configure file.

• tried to use the generated configure file to auto install and it failed at Cannot accommodate the required filesystem for installation. the job is https://openqa.suse.de/tests/16156378

• I've cut the screenshot for the error message.

• So @JERiveraMoya is there any setting missing for my configure file. Thanks

hjluo wrote in #note-8:

• So @JERiveraMoya is there any setting missing for my configure file. Thanks

Please, use Slack channel for technical doubts and discussion.
(you need to create the profile for latest version as recommended in the ticket description)
For ticket directions or guidance is fine if you use progress.

OK, I'll turn to slack for discussion. and I've generated profile as description in this job. thanks

• Now try to generate autoyast profile from yast lvm on 15sp7 case and compare it with the json profile from agama manual test.

1. Today tried to run default_sle with MULTIPATH=1 when activate pages shows click Yes and wait the installation to finish. then switched to root-console and found that agama config show output same storage.
2. asked the dev and they said can provide a workaroud for answer file.

According to developers we should include in logs the content of /run/agama/scripts attached in openQA.
Let's try to include this in this ticket if possible. Take into account that it shouldn't fail if the folder is empty (for other installation is empty).

JERiveraMoya wrote in #note-13:

According to developers we should include in logs the content of /run/agama/scripts attached in openQA.
Let's try to include this in this ticket if possible. Take into account that it shouldn't fail if the folder is empty (for other installation is empty).,

You mean to collect the log from /run/agama/scripts output? I'd try to collect that log.

hjluo wrote in #note-14:

JERiveraMoya wrote in #note-13:

According to developers we should include in logs the content of /run/agama/scripts attached in openQA.
Let's try to include this in this ticket if possible. Take into account that it shouldn't fail if the folder is empty (for other installation is empty).,

You mean to collect the log from /run/agama/scripts output? I'd try to collect that log.

yes, it is not a log file, they are referring to grab all content of that folder, it should be some example in openQA to borrow.

JERiveraMoya wrote in #note-15:

hjluo wrote in #note-14:

JERiveraMoya wrote in #note-13:

According to developers we should include in logs the content of /run/agama/scripts attached in openQA.
Let's try to include this in this ticket if possible. Take into account that it shouldn't fail if the folder is empty (for other installation is empty).,

You mean to collect the log from /run/agama/scripts output? I'd try to collect that log.

yes, it is not a log file, they are referring to grab all content of that folder, it should be some example in openQA to borrow.
got it, thanks a lot.

• Asked dev and use the pre script to active mutltipath service as workaround.
• Now multipathd.socket and multipathd [all activated] before probing. (https://openqa.suse.de/tests/16203240#step/agama_auto/17) at the end of the installation, but multipath -l gave me nothing while agama config show shows /dev/vda was used for disk
• from Knut Anderssen

so, probably the problem is that the probing is arleady done when multipath is activated.
I should finish current PBI which should solve the issue definitely as well as the profile should support a multipath activation itsef

• Knut Anderssen found 1 errro in the configure file and send me a new pre script.

"scripts": {
      "pre": [
         {
            "body": "#!/bin/bash\nif ! systemctl status multipathd ; then\n  echo \"Activating multipath\"\n  systemctl start multipathd.socket\n  systemctl start multipathd\n  sleep 1\nfi\n",
            "name": "activate multipath"
         }
      ]
   },

but after installation it shows multipath not found

hjluo wrote in #note-18:

but after installation it shows multipath not found

Job is missing MULTIPATH=1.

pcervinka wrote in #note-19:

hjluo wrote in #note-18:

but after installation it shows multipath not found

Job is missing MULTIPATH=1.

OK, thanks, I'm think of the same idea with compare to the manual test. will try to verify it.

• PR: https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/20867

• Now after PO is back,is in code review
• works for aarch with HDDMODEL=scsi-hd https://openqa.suse.de/tests/16353511

PR merged, please setup in corresponding job groups (dev and prod).

JERiveraMoya wrote in #note-25:

PR merged, please setup in corresponding job groups (dev and prod).

added sles_default_multipath_unattended for group 607 and 612.

hjluo wrote in #note-26:

JERiveraMoya wrote in #note-25:

PR merged, please setup in corresponding job groups (dev and prod).

added sles_default_multipath_unattended for group 607 and 612.

As the build do not trigger much often in prod, when I mean "add" I'm referring to [add + trigger] them in both job groups so we can have a clear view. Thanks.

multipath is already specific, doesn't need *_default_* in the test suite name, could you please drop that?

JERiveraMoya wrote in #note-28:

multipath is already specific, doesn't need *_default_* in the test suite name, could you please drop that?

Dropped, but I checked out in prod (612) looks like the pre script is not support as we expected.

hjluo wrote in #note-29:

JERiveraMoya wrote in #note-28:

multipath is already specific, doesn't need *_default_* in the test suite name, could you please drop that?

Dropped, but I checked out in prod (612) looks like the pre script is not support as we expected.

neither hashed password or scc yet, prod is behind in features than development, could you try without hashed password, I'm wondering if that makes it fails in the first places and does't process the pre script ?

could you also rename with quick PR the jsonnet profile? sle_default_multipath.jsonnet
what is missing in development? https://openqa.suse.de/tests/16362774/

JERiveraMoya wrote in #note-31:

could you also rename with quick PR the jsonnet profile? sle_default_multipath.jsonnet
OK. https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/20900

what is missing in development? https://openqa.suse.de/tests/16362774/
This case sometimes will failed at ask for password this morning. but rerun passed.

JERiveraMoya wrote in #note-30:

hjluo wrote in #note-29:

JERiveraMoya wrote in #note-28:

multipath is already specific, doesn't need *_default_* in the test suite name, could you please drop that?

Dropped, but I checked out in prod (612) looks like the pre script is not support as we expected.

neither hashed password or scc yet, prod is behind in features than development, could you try without hashed password, I'm wondering if that makes it fails in the first places and does't process the pre script ?

Just tried and it works with plain password. so should we set one profile for prod or all use plan password? thanks

Answering your questions:

• Run x10 for stability and report problems.
• Duplicate files with using similar file name convention than in existing solution for other test suites.

JERiveraMoya wrote in #note-34:

Answering your questions:

• Run x10 for stability and report problems.
• Duplicate files with using similar file name convention than in existing solution for other test suites.

• Just run 10 jobs and 3 of 10 jobs will hang at password screen. thanks
• asked in help-installation-sle16

• the dev said it's a known bug 1235022 with repository timeout.
• > Hi, it is known issue with repository timeout. We will need better way to handle it in Agama.

I move it here to avoid overlap this message:

• Just tried in prod and it works with plain password. Thanks

• so for now, how can we handle plain password in prod group? thanks

hjluo wrote in #note-38:

• so for now, how can we handle plain password in prod group? thanks

Please use Slack in general for your questions, I don't know how much info do you need, and having conversations here is quite ineffective, progress can be used for report 'a bit', decision points, blockers, but Slack is much better tool. Also there you can get support from others, in particular your Scrum Master, because I am the one that read the tickets more to answer, so it might also a bit of a bottle neck to ask too many questions here, I hope it makes sense for you. It will help to make things smooth I believe.

You already have an answer for this in https://progress.opensuse.org/issues/173659#note-34 I believe.

• Now marked failed jobs in prod with this poo, cause in prod group, we are not support pre/post script by now.

• PR to add profile to prod group: https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/20917