March 2022 - openSUSE Build Service - openSUSE Mailing Lists

Server side services running for a loooong time, even if they are of type "disabled"?
by Johannes Kastl 04 Mar '22

04 Mar '22

Hi all, I just committed two packages to OBS, both containing a _service file with only services of type "disabled". https://openbuildservice.org/help/manuals/obs-user-guide/cha.obs.source_ser… For some reason both of them are now running the services on server side and are blocked until those are through. Did not happen for similar packages in the past, so I am puzzled why that happens. Is that expected or a hickup in OBS? Kind regards, Johannes -- Johannes Kastl Linux Consultant & Trainer Tel.: +49 (0) 151 2372 5802 Mail: kastl(a)b1-systems.de B1 Systems GmbH Osterfeldstraße 7 / 85088 Vohburg http://www.b1-systems.de GF: Ralph Dehner Unternehmenssitz: Vohburg / AG: Ingolstadt,HRB 3537

2 2

OBS Image builds using device mapper modules broken in Tumbleweed
by Marcus Schäfer 04 Mar '22

04 Mar '22

Hi, We maintain a bunch of integration test builds in obs for testing kiwi features. There are integration tests builds also for Tumbleweed which recently stopped working when they use device mapper features as in luks or raid. Please see the following buildlogs: osc rbl Virtualization:Appliances:Images:Testing_x86:tumbleweed test-image-luks images x86_64 --- [ 289s] [ 266.952479][ T4505] device-mapper: table: 254:3: crypt: Error allocating crypto tfm (-ENOENT) --- or the raid test osc rbl Virtualization:Appliances:Images:Testing_x86:tumbleweed test-image-raid images x86_64 --- [ 90s] modprobe: FATAL: Module md_mod not found in directory /lib/modules/5.16.10-1-default --- For reference all this still works on leap e.g osc rbl Virtualization:Appliances:Images:Testing_x86:leap test-image-luks images x86_64 The testing of image features using device mapper modules or capabilities has stopped to work one or two weeks ago. I assume the kernel has changed in TW and as kernel-obs-build is just a flavor of kernel-default it received all the changes. I'm just guessing, but as of now you will not be able to create images for TW including this sort of features. It would be nice if this can be fixed in the kernel used by obs to start the workers. Neither support: kernel-obs-build nor support: kernel-default has an effect. I assume there is not much I can do to influence this ? Thanks a ton Cheers, Marcus -- Public Key available via: https://keybase.io/marcus_schaefer/key.asc keybase search marcus_schaefer ------------------------------------------------------- Marcus Schäfer Am Unterösch 9 Tel: +49 7562 905437 D-88316 Isny / Rohrdorf Germany -------------------------------------------------------

4 11

EPEL9 Project broken?
by Stefan Bluhm 04 Mar '22

04 Mar '22

Good Morning Adrian, can you please have a look at the Fedora:EPEL project? Specifically the EPEL:9 project? It appears to be broken. New packages are not appearing for quite some time. And there are red triangles in the status: https://build.opensuse.org/project/show/Fedora:EPEL:9 Linked repo URL does work and packages are there. Thank you and best wishes, Stefan

2 3

Resigning every package and repository in a private OBS?
by Neal Gompa 01 Mar '22

01 Mar '22

Hey all, Is there a way to trigger OBS to resign every package and repository? It looks like I need to do this for two things: upgrade the GPG key to a stronger algorithm for APT[1] and to change the hash algorithm to SHA256 in obs-signd[2] because CentOS/RHEL 9 no longer accepts SHA1 by default[3]. [1]: https://wiki.debian.org/Teams/Apt/Sha1Removal [2]: https://github.com/openSUSE/obs-sign/issues/34 [3]: https://bugzilla.redhat.com/show_bug.cgi?id=2031742 -- 真実はいつも一つ！/ Always, there's only one truth!

1 0

Signer performance
by Kai Liu 01 Mar '22

01 Mar '22

Hi, It seems all gpg processes signers use, are talking to one single gpg-agent, and it only can make use of one core? Is there a way to make the signing process more "parallel"? (without adding more servers...) Regards, Kai

1 0