1. Forum
  2. Usenet
  3. LINUX.DEBIAN.DEVEL

  • Re: How should Debian CI handle autopkgtests that require more than thr

    From Antonio Terceiro@21:1/5 to Soren Stoutner on Fri May 9 01:50:01 2025
    On Thu, May 08, 2025 at 11:38:35AM -0700, Soren Stoutner wrote:
    I have two questions.

    1. Is there a way to override the 3 hour time limit for Debian CI for a particular package?

    No.

    2. Would there be objections to reconsidering the 3 hour default time limit for all packages?

    We could allow longer timeouts on a per-package basis and on maintainer request, like we handle having packages tested with KVM instead of
    containers.

    This would need the code to actually support this to be written first,
    though. We need to be careful with this to prevent packages with
    long-running test suites from causing a DoS on the infrastructure.

    I would hate to have to disable autopkgtest for some or all architectures on these packages, as the output has been beneficial. But I also don’t want to
    prevent migration to testing just because the tests are timing out.

    Yes, I agree.

    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCAAdFiEEst7mYDbECCn80PEM/A2xu81GC94FAmgdQRkACgkQ/A2xu81G C94daw/9G5zsf6YTp3ov4xbxrmxW4XXT36rZ/RnwHyoE/yhOJsNWdzY1qM5Xczlq 58/HYSrQRZQhbEmFcdz1V+OV0bCr+8V7VEX18pIwUC6XCzmLI5lH27JL81ZalESr Lyl7y/HQDXe6FD0f9/JeKZ3hEQ+Xg7QdTQlkH4OfeZiLP5Hu9Js6GxjXIxyIytDz 4ZpSdjfT5rWJT3Cv/Qas5iSVSZTwtriM21yWrepwG9ZwvW0Vkesv9XuTdWeX9m/H Yd+7rj9CvtMiVUW+BS7s3Dm7RBAZcab+OujORYzbkRac2WLMLOeoL53VCYnP331v 3JUlc5Oqs5pK4FehxUM8Mkxw/1IAJcJGZbx/vtHntIsMWMMaBlRXSEzhhhq/HPrm iQLc6NCXZNAI7QJBPn7IUrUJ8ZtqYmQ5/821pBcYwsk3RW/dm8jLB5tyJBBKU71g iZpZHka82eC8JplWyh4bLreJNZsTFnmQ1DTYfHfYOVttOS//uU+DIt0gmvMSIA9Y aOllLbnR+Ac4xh5dJMdo5hbGwpQ8t6yup8DdH48MIHnKwa4xNYYgOvdl2KHHFV2e nN11LuGmK1/mIlGMtI08HJdb5EpGgGBL/MAHqfwK9fKyELqo5cLTa5FlRlX2lP47 HspZXIB6L760CrBngdrRzQAtijn3K/IBKSjLOUm+1NCGyKq5Gww=
    =vFxt
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Soren Stoutner@21:1/5 to All on Thu May 8 18:28:18 2025
    On Thursday, May 8, 2025 4:41:16 PM Mountain Standard Time Antonio Terceiro wrote:
    We could allow longer timeouts on a per-package basis and on maintainer request, like we handle having packages tested with KVM instead of containers.

    This would need the code to actually support this to be written first, though. We need to be careful with this to prevent packages with
    long-running test suites from causing a DoS on the infrastructure.

    I agree. Being able to allow longer timeouts on a per-package basis sounds like a good solution. Who would be the correct people to talk to about enabling such functionality?

    --
    Soren Stoutner
    soren@debian.org
    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEEJKVN2yNUZnlcqOI+wufLJ66wtgMFAmgdWjIACgkQwufLJ66w tgP+YRAApB1/5v3Ft2Zy+aYzuiUsQ5y/BxqWP52N3heyN6dD8oOgIQR3FH7/FRRK /Et2p4KyGus8WlxKJ0Xy6E9wr0u4gaiGqGV0WZzPg1cFACQcCQYrPweQHNx4o63d 0j4XbTcjhl5uO6k+j3XF6++JeoitSyw1qxKrekC3kXizTWQos5KvuqMrR3y3uahK JdMFWMjfjnVhXDFqc6fVz6y5GmytQUL7Ec0dduQsJoTSd9sXqhefPnN4+iIxLZHY k1mjwXKz9Y/7/AIoEFt1BBuM1EvfBzQ6ZPwJYSidgzIlI1OnZI8dPz1iJjtWw/F3 o+opOYWYOYIyRwUDkKtaZOg4pcAOxJ44UYR/+8YYfvqDEiHLVzwyizlq78m3kiG8 MRfCLP1cajU4eZrEW5Ce7i/gcT6BqHbUF1ZzHhO/WPT/MNLhVL4eAB32b03NiBdn 7RsN6Yfc8msCqTisUvFbns3JA3u0+CMmobiSd+vvoWABx28ehtT8wdjt11Xt2ryM e4HsfWIHn5rgRhDITnZDdiAUPWwdrdK98LI7nuXugul9d6zGv9NZAe6mibr6u/4w m2+pjUS9fEAlZes/5hh8Cc9uGQKNakRmMI69k/nyr0Wpkw3c84EYgceaqEImmh0w ij1iiKU7a+IzVTdf5W4KNv+JcjwJzzafrD91jKvLqUxcJVrWnTI=
    =K9ia
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Paul Gevers@21:1/5 to All on Fri May 9 08:40:01 2025
    This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --------------alIXZRKsmHWIovbUnQ3zgYW0
    Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: base64

    SGksDQoNCk9uIDA5LTA1LTIwMjUgMDM6MjgsIFNvcmVuIFN0b3V0bmVyIHdyb3RlOg0KPiBP biBUaHVyc2RheSwgTWF5IDgsIDIwMjUgNDo0MToxNuKAr1BNIE1vdW50YWluIFN0YW5kYXJk IFRpbWUgQW50b25pbyBUZXJjZWlybw0KPiB3cm90ZToNCj4+IFdlIGNvdWxkIGFsbG93IGxv bmdlciB0aW1lb3V0cyBvbiBhIHBlci1wYWNrYWdlIGJhc2lzIGFuZCBvbiBtYWludGFpbmVy DQo+PiByZXF1ZXN0LCBsaWtlIHdlIGhhbmRsZSBoYXZpbmcgcGFja2FnZXMgdGVzdGVkIHdp dGggS1ZNIGluc3RlYWQgb2YNCj4+IGNvbnRhaW5lcnMuDQo+Pg0KPj4gVGhpcyB3b3VsZCBu ZWVkIHRoZSBjb2RlIHRvIGFjdHVhbGx5IHN1cHBvcnQgdGhpcyB0byBiZSB3cml0dGVuIGZp cnN0LA0KPj4gdGhvdWdoLiBXZSBuZWVkIHRvIGJlIGNhcmVmdWwgd2l0aCB0aGlzIHRvIHBy ZXZlbnQgcGFja2FnZXMgd2l0aA0KPj4gbG9uZy1ydW5uaW5nIHRlc3Qgc3VpdGVzIGZyb20g Y2F1c2luZyBhIERvUyBvbiB0aGUgaW5mcmFzdHJ1Y3R1cmUuDQo+IA0KPiBJIGFncmVlLiAg QmVpbmcgYWJsZSB0byBhbGxvdyBsb25nZXIgdGltZW91dHMgb24gYSBwZXItcGFja2FnZSBi YXNpcyBzb3VuZHMNCj4gbGlrZSBhIGdvb2Qgc29sdXRpb24uICBXaG8gd291bGQgYmUgdGhl IGNvcnJlY3QgcGVvcGxlIHRvIHRhbGsgdG8gYWJvdXQNCj4gZW5hYmxpbmcgc3VjaCBmdW5j dGlvbmFsaXR5Pw0KDQoNCkFuIGFsdGVybmF0aXZlIHRoYXQgY2FuIGJlIGRvbmUgbm93OiB0 aGUgdGltZW91dCBvZiAyOjQ3aCBpcyBwZXIgdGVzdCANCnN0YW56YSBpbiBkL3QvY29udHJv bCAoc2VlIGF1dG9wa2d0ZXN0KDEpLiBGb3Igc29tZSB0ZXN0IHN1aXRlcyBpdCdzIG5vdCAN CmhhcmQgdG8gc3BsaXQgdGhlIHRlc3Qgc3VpdGUgb3ZlciBtdWx0aXBsZSBzdGFuemEuIFRo ZSBvdmVyYWwgdGltZW91dCBpcyANCjguNWggKGV4Y2VwdCBvbiByaXNjdjY0IHdoZXJlIGl0 J3MgY3VycmVudGx5IGRvdWJsZWQpLg0KDQpQYXVsDQo=

    --------------alIXZRKsmHWIovbUnQ3zgYW0--

    -----BEGIN PGP SIGNATURE-----

    wsB5BAABCAAjFiEEWLZtSHNr6TsFLeZynFyZ6wW9dQoFAmgdoU0FAwAAAAAACgkQnFyZ6wW9dQo6 rAf7BYAj+KQWk5bppvy8aH/W6dW5NcaVf+YEnscCem1U7TU7U10MxE75j/6j9PnMEWdIY2L/G50Z 7RTCEb5Jvb/Ou2Qc6+U0Q+mDLUQVxJ3U+vvVYXYwfdtFyRJa9xqzm8Klc8Cb2ocbKpQgt5AS1YXP EQy2fMtJt3tED28A9vX0w7L1oInDiSW0Os3rvkhZHN7UDHHqZJG0dAVVSmmCLOwDQZl57JPZeop2 wzWQX5+ZFM38ft5ngw+WF/8CMmq+nsbJVHzvjh1KDStcV2pDkvGio8Hbvq9eAxE6a3GwJjykZSup uOOMx+8zdz0Y0MxRPTfditm4LgpH+G1PjoTZOQSw0Q==
    =6bvS
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Soren Stoutner@21:1/5 to All on Fri May 9 11:05:02 2025
    On Thursday, May 8, 2025 11:31:41 PM Mountain Standard Time Paul Gevers wrote:
    An alternative that can be done now: the timeout of 2:47h is per test
    stanza in d/t/control (see autopkgtest(1). For some test suites it's not hard to split the test suite over multiple stanza. The overal timeout is 8.5h (except on riscv64 where it's currently doubled).

    Thank you for that information. I did not know that there were separate timeouts for individual stanzas vs. the entire test suite. I think I should be able to split my tests into separate stanzas and I expect the overall timeout of 8.5h should be sufficient for anything I am currently working on.

    There might be other packages for which this is still not sufficient, but for me I don’t think any other adjustments are necessary.

    --
    Soren Stoutner
    soren@debian.org
    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEEJKVN2yNUZnlcqOI+wufLJ66wtgMFAmgeQ84ACgkQwufLJ66w tgO7zQ//cflWn16dAQZicXSS25lqvW+xfY1EmhSIkKaDPRuOwbO3KVjN5UuXlK6y hRXCCyh32MKqE4PX0vb2R4PLNRPqKdPsamxAGgQqwgvKC2P2hhSZ6Lcsge+FRGJU 8yMVrWrdvkNQAITdTaCAjYHD3PCdGybO0d8ZwRK47FtxRk2u1GhoRKZtUj+Dfkd4 cTqjYDSGzdsHeMkybzYs5oPdpVZf9GnoorCgZdJ5fcG/MfuepF8VX9WfUT7oEDI7 ayxZNzYdGkchfyNXOw9OSTOzsl0sbyh028tQ734ZT5+8NTgYvQlPzaiCO0be/Jfr y/6MvkxT2Sdt/Jd2+9ku1UWDtWtAqAH3OlPrAunqo2DbOnz3KIFnJgyVv9cn5nwi IiLSUev/6OZ+09nzCa3gJVGcb1S5JTOvBrVRHjcgLctMN2liTvQZ5oSe6X2JSc/T x3ubHhEQ9njivMdbf+i6cDPxgaLKO0z/pR8mHs6E6b6NuXrp3qAzKf0HaQdwp7Yp vXSPR5YABNuE/8IOSFRp6yICE66rO7EkdXQzaeRTSEhYcaIH5Q6dLhqHj6eLBAMV nf9PBpl/rSRnEdNnRysyUrrmYsoluc2HQGRGZgdrc+s0lBg446n7si7FYBchi1uV AlNhJFBbWEN30PD4K6k22RR1A4U/7E8lX4t4KJRI7uwLmJO6nQQ=
    =3KT7
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)