Bugzilla – Bug 1219917
rasdaemon is missing dmidecode dependency on ppc64le
Last modified: 2024-04-16 13:41:55 UTC
## Observation openQA test in scenario opensuse-Tumbleweed-DVD-ppc64le-extra_tests_kernel@ppc64le fails in [rasdaemon](https://openqa.opensuse.org/tests/3937353/modules/rasdaemon/steps/3) ## Test suite description Maintainer: kernel-qe@suse.de. Extra tests group for non-LTP kernel tests. ## Reproducible Fails since (at least) Build [20240207](https://openqa.opensuse.org/tests/3924081) ## Expected result Last good: [20240206](https://openqa.opensuse.org/tests/3921430) (or more recent) ## Further details Always latest result in this scenario: [latest](https://openqa.opensuse.org/tests/latest?arch=ppc64le&distri=opensuse&flavor=DVD&machine=ppc64le&test=extra_tests_kernel&version=Tumbleweed) We recently enabled rasdaemon test on more architectures in openQA. rasdaemon fails during installation on ppc64le, it is missing dmidecode. Loading repository data... Reading installed packages... Resolving package dependencies... Problem: nothing provides 'dmidecode' needed by the to be installed rasdaemon-0.8.0.39.git+cfabd93-1.2.ppc64le Solution 1: do not install rasdaemon-0.8.0.39.git+cfabd93-1.2.ppc64le Solution 2: break rasdaemon-0.8.0.39.git+cfabd93-1.2.ppc64le by ignoring some of its dependencies It's clear that dmidecode is not available for ppc64le. There was even similar issue for SLE years ago, but seems to be fixed for s390x only. See https://bugzilla.opensuse.org/show_bug.cgi?id=1071830 rasdaemon spec file contains: %ifnarch s390x Requires: dmidecode %endif Not sure what should be correct state, above exception should be extended to ppc64le as well or don't ship rasdaemon. Just note rasdaemon is not shipped on ppc64le in SLE.
Excluding dmidecode from Requires list for rasdaemon at least for Tumbleweed should be fine -> will do.
Hello SUSE, I have mirrored this bug to IBM LTC Bug 205828. The IBM team update about it will show up here. Thank you for your support.
Thanks to Michal Suchanek this is in Factory now: https://build.opensuse.org/request/show/1156945
(In reply to Thomas Renninger from comment #4) > Thanks to Michal Suchanek this is in Factory now: > https://build.opensuse.org/request/show/1156945 Hello Petr, Could you please confirm here once you will have verified this bug? Thank you for your support.
(In reply to Géry Schneider from comment #6) Hi Gery, > Hello Petr, > Could you please confirm here once you will have verified this bug? > Thank you for your support. I can confirm that installation if fine now: Retrieving: rasdaemon-0.8.0.49.git+f9cb13b-1.1.ppc64le.rpm [.done] Checking for file conflicts: [.....done] (1/8) Installing: libtraceevent1-1.7.3-1.3.ppc64le [...done] (2/8) Installing: libxcrypt-devel-4.4.36-1.3.ppc64le [..done] (3/8) Installing: linux-glibc-devel-6.8-1.1.ppc64le [...done] (4/8) Installing: perl-DBI-1.643-3.1.ppc64le [..done] (5/8) Installing: glibc-devel-2.39-4.1.ppc64le [...done] (6/8) Installing: sqlite3-devel-3.45.2-1.1.ppc64le [..done] (7/8) Installing: perl-DBD-SQLite-1.740.0-2.1.ppc64le [..done] (8/8) Installing: rasdaemon-0.8.0.49.git+f9cb13b-1.1.ppc64le [.. Updating /etc/sysconfig/rasdaemon ... done] Running post-transaction scripts [....done] But our tests fails further on command: # ras-mc-ctl --mainboard Missing command in piped open at /sbin/ras-mc-ctl line 424. https://openqa.opensuse.org/tests/4045378#step/rasdaemon/20 We never run rasdaemon test on ppc64le, so I'm not sure what should work or not. Should I file new bug?
------- Comment From sthoufee@in.ibm.com 2024-04-01 01:08 EDT------- We got following response from RAS Developement team. RAS team has no plans to own or support this tool. rasdaemon is specific to x86 platform and may not work on ppc64 at all. Powerpc has its own RAS tools which are shipped through following packages and available at https://github.com/power-ras/ ppc64-diag lsvpd servicelog rasdeamon mostly deals with MCE logs. On Power, for memory errors, predictive error analysis is done by pHyp/PRD as similar to rasdaemon on x86 platform. pHyp then delivers these predictive errors to linux as RTAS events which captured by rtas_errd daemon and logs into servicelog as serviceabile events (Call home event). We should check with SuSE on purpose of this tool on ppc64 and why do they think this will work on ppc64. Do we really need this tool on ppc64 ? Who is going to own/support this on ppc64 ?
It is true that the primary feature of rasdaemon should be handled by ppc64-diag. However, rasdaemon grew other features over time, and it is not clear that none are applicable on powerpc.
I'm sorry, but it is quite confusing. Should we still run rasdaemon test on ppc64le in TW? If no, why it is included as package?
@Petr: I would not add any ppc64le rasdaemon tests. I would see adding ppc64le as a possibility for others to play around with it on ppc64le.
Thank you @Thomas, that helps. I will update the test to do just installation on tw@ppc64le without any "functional" test.
Installations works fine: https://openqa.opensuse.org/tests/4082698#step/rasdaemon/3
------- Comment From gery.schneider@fr.ibm.com 2024-04-16 09:38 EDT------- (In reply to comment #22) > Installations works fine: > https://openqa.opensuse.org/tests/4082698#step/rasdaemon/3 Thank you Petr for this verification, and all for your support. As per the comments above, I am closing this bug.