Univention Bugzilla – Full Text Bug Listing |
Summary: | [Performance] App Center overview | ||
---|---|---|---|
Product: | UCS Test | Reporter: | Stefan Gohmann <gohmann> |
Component: | App Center | Assignee: | Jürn Brodersen <brodersen> |
Status: | CLOSED FIXED | QA Contact: | Alexander Kläser <klaeser> |
Severity: | normal | ||
Priority: | P5 | CC: | klaeser, wiesenthal |
Version: | unspecified | ||
Target Milestone: | --- | ||
Hardware: | Other | ||
OS: | Linux | ||
What kind of report is it?: | --- | What type of bug is this?: | --- |
Who will be affected by this bug?: | --- | How will those affected feel about the bug?: | --- |
User Pain: | Enterprise Customer affected?: | ||
School Customer affected?: | ISV affected?: | ||
Waiting Support: | Flags outvoted (downgraded) after PO Review: | ||
Ticket number: | Bug group (optional): | ||
Max CVSS v3 score: | |||
Bug Depends on: | 39632, 40239, 40240 | ||
Bug Blocks: |
Description
Stefan Gohmann
2015-05-15 07:12:20 CEST
I added a performance test for the UMC App Center. The current timeout after which the test fails is set to 5 sec. UCS4.1-0: r65118: * Added performance test for the appcenter (Bug #38545) Package: ucs-test Version: 6.0.12-11.1316.201511031208 Superb :) ! As the test currently fails, it would be nice to mark the test case as known such that the test will not be executed at the moment. Please create a new bug which states that the performance for the query command needs to be improved. UCS4.1-0: r65232: * Set 20_appcenter/60_check_appcenter_performance to SKIP (Bug #38545; #39632) Package: ucs-test Version: 6.0.13-2.1322.201511051351 Current results are as follows... not sure whether the test should be re-activated. The first round probably syncs app center data. @Dirk: What do you think? > root@ucs-dc-4621:~# /usr/share/ucs-test/20_appcenter/60_check_appcenter_performance > ### FAIL ### > The appcenter answered too slow > Threshold is 5 sec; Appcenter replied in 12.5546138287 sec. > ### ### > Starting 1 ucs-test at 2015-12-22 04:35:09 to /dev/null > Check the performance for requests against the UMC appcenter..................................................................... Test failed > > root@ucs-dc-4621:~# /usr/share/ucs-test/20_appcenter/60_check_appcenter_performance > Starting 1 ucs-test at 2015-12-22 04:35:20 to /dev/null > Check the performance for requests against the UMC appcenter..................................................................... Test passed > > root@ucs-dc-4621:~# /usr/share/ucs-test/20_appcenter/60_check_appcenter_performance > Starting 1 ucs-test at 2015-12-22 04:35:35 to /dev/null > Check the performance for requests against the UMC appcenter..................................................................... Test passed I could change the test to ignore the first app center call. But I think a long loading time for the first app center start could give a bad first impression. Would it be possible to sync the app center data as part of the system-setup or with a cron job? (In reply to Jürn Brodersen from comment #5) > I could change the test to ignore the first app center call. But I think a > long loading time for the first app center start could give a bad first > impression. > > Would it be possible to sync the app center data as part of the system-setup > or with a cron job? See Bug 39632, comment #10: > (In reply to Alexander Kläser from comment #9) > > I think a cron job would be nice to avoid syncing the data which leads to > > bad performance with the first request. > > Actually, there is a cron job that does that, though not directly > univention-app update. > > But it is called as part of univention-upgrade which runs once a day. What about calling univention-app update before running the test? This should be a fair performance test then. IMHO it cannot really be forseen how long the synchronization of the App Center data will take. The app center is now updated before the request is made. UCS4.1-0: r68072: * 20_appcenter/60_check_appcenter_performance: Update Appcenter before testing the appcenter performance Package: ucs-test Version: 6.0.33-42.1448.201603141510 OK for now. The test case fails from time to time: http://jenkins.knut.univention.de:8080/job/UCS-4.1/job/UCS-4.1-4/job/AutotestJoin/SambaVersion=s3,Systemrolle=member/111/testReport/20_appcenter/60_check_appcenter_performance/test/ [2017-02-20 21:36:43.714607] Appcenter update starting [2017-02-20 21:36:52.773051] Appcenter update done [2017-02-20 21:36:52.775495] Getting umc_connection [2017-02-20 21:36:53.126296] Start request to appcenter [2017-02-20 21:37:05.261161] Request finished [2017-02-20 21:37:05.261288] ### FAIL ### [2017-02-20 21:37:05.261332] The appcenter answered too slow [2017-02-20 21:37:05.261348] Threshold is 4.5 sec; Appcenter replied in 4.58675909042 sec. [2017-02-20 21:37:05.261380] ### ### Should I change the timeout or deactivate the test? (In reply to Jürn Brodersen from comment #10) > Should I change the timeout or deactivate the test? I would suggest to change the timeout to 5sec (or a bit higher) then, although I would consider this behaviour to be a bug. 5sec really is too slow. (In reply to Alexander Kläser from comment #11) > (In reply to Jürn Brodersen from comment #10) > > Should I change the timeout or deactivate the test? > > I would suggest to change the timeout to 5sec (or a bit higher) then, > although I would consider this behaviour to be a bug. 5sec really is too > slow. Yes, let's try 5 seconds and yes I also think it is too slow. I am working on App Setting. This may speedup the process, maybe we can keep up the threshold. r77089: 20_appcenter/60_check_appcenter_performance: increase timeout Package: ucs-test Version: 6.0.37-56.1598.201702240948 Branch: ucs_4.1-0 Scope: errata4.1-4 I increased the timeout for now. (In reply to Dirk Wiesenthal from comment #13) > I am working on App Setting. This may speedup the process, maybe we can keep > up the threshold. In the meantime, the threshold was set to 15 sec. From one of the last runs: > Success: The appcenter answered in 8.87774896622 sec. @Dirk: This really is not acceptable! Nevertheless, this bug can be set to VERIFIED. |