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

As mentioned in daily (was a comment for other colleagues afir), to get the selector is a process that we might need to pair up a bit.
Please, try to focus on having the PR ready even with 'weak' selectors and then we will help a bit with those as a team.

• PR: https://github.com/jknphy/agama-integration-test-webpack/pull/85

hjluo wrote in #note-7:

Now in Build 57.3 the screen changed, there's no kvm_host just fips, selinux and cockpit. and I've checked software_selection_page.ts and it don't fit current page.
now need adapt the new page with software_selection_page.ts.

57.3 has a lot of patterns to be installed included the one for kvm, in this ticket we should provide the functionality to select as many as we want via UI.

PR to validate the installed patterns

• https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/21394

• Now fixed issue for some patterns were not selected after click, the rest job is to run O3 cases to make sure there's no regression, but today was blocked by O.S.D not work forticket_179317.

• Now Puppeteer PR#85 was merged.
• Changed O3 group to use test_software_selection
• Will add new test case sles_software_selection for dev and prod.

• Added case sles_software_selection to prod group and trigged
• Added case sles_software_selection to dev group and was auto triggered.

hjluo wrote in #note-12:

• Added case sles_software_selection to prod group and trigged
• Added case sles_software_selection to dev group and was auto triggered.

For your PR verification runs, I would expect you select a lot of patterns, they might vary among architectures, but let's select as many as possible. For now I see some problems in the automation. Please label the failure with your ticket if it is the case.

• Now test with all patterns that listed on the page. For 66.1 they are:

'documentation,kvm_tools,selinux,sw_management,file_server,mail_server,
lamp_server,gateway_server,dhcp_dns_server,directory_server,kvm_server,
fips,cockpit'

• Checked the auto results, for the failures,the reasons are:

  • 1. we need scrollIntoView if we have lots of pattern to select.
  • 2. We need to check if it's auto-selected before click.

• I've addressed these issues with a PR#93 to fix it and the results looks good with all 13 patterns selected.

hjluo wrote in #note-14:

• Now test with all patterns that listed on the page. For 66.1 they are:

'documentation,kvm_tools,selinux,sw_management,file_server,mail_server,
lamp_server,gateway_server,dhcp_dns_server,directory_server,kvm_server,
fips,cockpit'

• Checked the auto results, for the failures,the reasons are:

  • 1. we need scrollIntoView if we have lots of pattern to select.
  • 2. We need to check if it's auto-selected before click.

• I've addressed these issues with a PR#93 to fix it and the results looks good with all 13 patterns selected.

can you share why you think 1) is needed ? I didn't see any issue with scrolling at the end.
regarding 2) as I told you before we don't need to check if they are auto-selected, because we are going to pass only patterns that are not auto-selected.
In other words, we want to be specific and fix expectations in our tests, if the product has a wrong auto-selected pattern we would not notice with your fix.

JERiveraMoya wrote in #note-15:

hjluo wrote in #note-14:

• Now test with all patterns that listed on the page. For 66.1 they are:

'documentation,kvm_tools,selinux,sw_management,file_server,mail_server,
lamp_server,gateway_server,dhcp_dns_server,directory_server,kvm_server,
fips,cockpit'

• Checked the auto results, for the failures,the reasons are:

  • 1. we need scrollIntoView if we have lots of pattern to select.
  • 2. We need to check if it's auto-selected before click.

• I've addressed these issues with a PR#93 to fix it and the results looks good with all 13 patterns selected.

can you share why you think 1) is needed ? I didn't see any issue with scrolling at the end.
regarding 2) as I told you before we don't need to check if they are auto-selected, because we are going to pass only patterns that are not auto-selected.
In other words, we want to be specific and fix expectations in our tests, if the product has a wrong auto-selected pattern we would not notice with your fix.

• This job w/o scrollIntoView failed.
• This job with scrollIntoView passed.
• I've removed the check for checked

yes, not sure, I didn't need the scroll (but I guess it doesn't hurt much) when trying in local.
Anyway it is merged, could you please adapt the patterns in production and development job group to install all those patterns?

• now changed PATTERNS to

PATTERNS:'documentation,kvm_tools,sw_management,file_server,mail_server,lamp_server,gateway_server,dhcp_dns_server,directory_server,fips,cockpit'

for prod and dev group and retriggered all arches.

• prod build 66.1
• for the failed ppc64le, looks we can't finished in 15 minutes. but my branch to wait 20 minutes finished in 13 minutes
• dev was blocked by import_agama_profile

hjluo wrote in #note-18:

• now changed PATTERNS to

PATTERNS:'documentation,kvm_tools,sw_management,file_server,mail_server,lamp_server,gateway_server,dhcp_dns_server,directory_server,fips,cockpit'

for prod and dev group and retriggered all arches.

• prod build 66.1
• for the failed ppc64le, looks we can't finished in 15 minutes. but my branch to wait 20 minutes finished in 13 minutes
• dev was blocked by import_agama_profile

yes, it makes sense that now we install so many package we would need more time, I adjusted installation recently to wait 14/15 minutes, 14 for the timeout of the action and 15 for the timeout of the test.

Could you do the same with values 19 and 20 and see if it fits. I would avoid to put much more, because we still didn't solve the issue of waiting for the error and early exit.

JERiveraMoya wrote in #note-19:

hjluo wrote in #note-18:

• now changed PATTERNS to

PATTERNS:'documentation,kvm_tools,sw_management,file_server,mail_server,lamp_server,gateway_server,dhcp_dns_server,directory_server,fips,cockpit'

for prod and dev group and retriggered all arches.

• prod build 66.1
• for the failed ppc64le, looks we can't finished in 15 minutes. but my branch to wait 20 minutes finished in 13 minutes
• dev was blocked by import_agama_profile

yes, it makes sense that now we install so many package we would need more time, I adjusted installation recently to wait 14/15 minutes, 14 for the timeout of the action and 15 for the timeout of the test.

Could you do the same with values 19 and 20 and see if it fits. I would avoid to put much more, because we still didn't solve the issue of waiting for the error and early exit.

Just tried 19 and 20 but the test finished in 8m 19s, so I think the network is the issue to take into account. so I rerun the failed job and it finished in 13 minutes. so we can hold it by now.