Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
Hi, sorry for the delay. Initially I wanted to include all tests in the packaging as they just didn't fail for me on ppc64el, amd64 and i386. Having a look at the .spec file upstream, they skip some of the tests, amongst them, the one we have an issue with. --- # don't require qemu within OBS # and exclude known flaky tests in OBS check # https://progress.opensuse.org/issues/52652 # 07-commands: https://progress.opensuse.org/issues/60755 for i in 07-commands 13-osutils 14-isotovideo 18-qemu-options 18-backend-qemu 99-full-stack; do rm t/$i.t done --- So I think we can safely skip them, if upstream does. I'm going to change this. Fred On Sat, 27 Mar 2021 17:12:17 +0200, Adrian Bunk wrote: > ping > > > On Fri, Mar 12, 2021 at 05:33:27PM +0200, Adrian Bunk wrote: > > On Thu, Feb 25, 2021 at 09:52:08AM +0100, Paul Gevers wrote: > > > Control: found -1 4.5.1527308405.8b586d5-4.2 > > > > > > Hi Frédéric, Hideki, > > > > > > On 17-02-2021 22:01, Paul Gevers wrote: > > > > If the forth time worked because of sheer luck, then please no, keep the > > > > bug open until the build is less flaky. We need packages to be build > > > > without failure [1]. Having to baby-sit flaky is not really an option as > > > > there are too many packages in the Debian archive. > > > > > > I had a look at the reproducible build project history for os-autoinst > > > [1] and the package FTBFS very, very often, both in unstable and > > > testing. I have marked this bug as found, so now this package is able to > > > migrate, *but* you'll have to fix this bug if you want the package to > > > ship with bullseye. If you can't fix the tests and still believe that > > > the package is in a good shape for the bullseye release, I suggest you > > > disable the tests for now. > > > > Or if the package is not in a good shape on i386, > > remove it from the architecture list. > > > > os-autoinst is amd64-only in buster. > > > > > Paul > > >... > > > > cu > > Adrian > signature.asc Description: PGP signature
Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
ping On Fri, Mar 12, 2021 at 05:33:27PM +0200, Adrian Bunk wrote: > On Thu, Feb 25, 2021 at 09:52:08AM +0100, Paul Gevers wrote: > > Control: found -1 4.5.1527308405.8b586d5-4.2 > > > > Hi Frédéric, Hideki, > > > > On 17-02-2021 22:01, Paul Gevers wrote: > > > If the forth time worked because of sheer luck, then please no, keep the > > > bug open until the build is less flaky. We need packages to be build > > > without failure [1]. Having to baby-sit flaky is not really an option as > > > there are too many packages in the Debian archive. > > > > I had a look at the reproducible build project history for os-autoinst > > [1] and the package FTBFS very, very often, both in unstable and > > testing. I have marked this bug as found, so now this package is able to > > migrate, *but* you'll have to fix this bug if you want the package to > > ship with bullseye. If you can't fix the tests and still believe that > > the package is in a good shape for the bullseye release, I suggest you > > disable the tests for now. > > Or if the package is not in a good shape on i386, > remove it from the architecture list. > > os-autoinst is amd64-only in buster. > > > Paul > >... > > cu > Adrian
Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
On Thu, Feb 25, 2021 at 09:52:08AM +0100, Paul Gevers wrote: > Control: found -1 4.5.1527308405.8b586d5-4.2 > > Hi Frédéric, Hideki, > > On 17-02-2021 22:01, Paul Gevers wrote: > > If the forth time worked because of sheer luck, then please no, keep the > > bug open until the build is less flaky. We need packages to be build > > without failure [1]. Having to baby-sit flaky is not really an option as > > there are too many packages in the Debian archive. > > I had a look at the reproducible build project history for os-autoinst > [1] and the package FTBFS very, very often, both in unstable and > testing. I have marked this bug as found, so now this package is able to > migrate, *but* you'll have to fix this bug if you want the package to > ship with bullseye. If you can't fix the tests and still believe that > the package is in a good shape for the bullseye release, I suggest you > disable the tests for now. Or if the package is not in a good shape on i386, remove it from the architecture list. os-autoinst is amd64-only in buster. > Paul >... cu Adrian
Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
Control: found -1 4.5.1527308405.8b586d5-4.2 Hi Frédéric, Hideki, On 17-02-2021 22:01, Paul Gevers wrote: > If the forth time worked because of sheer luck, then please no, keep the > bug open until the build is less flaky. We need packages to be build > without failure [1]. Having to baby-sit flaky is not really an option as > there are too many packages in the Debian archive. I had a look at the reproducible build project history for os-autoinst [1] and the package FTBFS very, very often, both in unstable and testing. I have marked this bug as found, so now this package is able to migrate, *but* you'll have to fix this bug if you want the package to ship with bullseye. If you can't fix the tests and still believe that the package is in a good shape for the bullseye release, I suggest you disable the tests for now. Paul [1] https://tests.reproducible-builds.org/debian/rb-pkg/bullseye/amd64/os-autoinst.html OpenPGP_signature Description: OpenPGP digital signature
Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
Hi Frédéric, On 15-02-2021 10:39, Frédéric Bonnard wrote: > indeed Paul, the test is flaky, probably a timing issue. I had already > changed some related parameter that made the tests pass reliably on > my i386 builder, but saw afterward that it still failed on debian's > builder and couldn't reproduce it : > https://salsa.debian.org/debian/os-autoinst/-/commit/38b2f4aefafbdf7ab59ce223ef208c540409a001 > I guess, we could increase CI related timeouts here and there but that's > what I tried to avoid when changing CI variable, and I was happy to see > things work.. but not on all i386 builders :) > May we close that bug for now ? If the forth time worked because of sheer luck, then please no, keep the bug open until the build is less flaky. We need packages to be build without failure [1]. Having to baby-sit flaky is not really an option as there are too many packages in the Debian archive. Also, your commit message suggests that the upstream change that's needed isn't in Debian yet? Paul [1] https://release.debian.org/bullseye/rc_policy.txt 4 OpenPGP_signature Description: OpenPGP digital signature
Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
Hi, indeed Paul, the test is flaky, probably a timing issue. I had already changed some related parameter that made the tests pass reliably on my i386 builder, but saw afterward that it still failed on debian's builder and couldn't reproduce it : https://salsa.debian.org/debian/os-autoinst/-/commit/38b2f4aefafbdf7ab59ce223ef208c540409a001 I guess, we could increase CI related timeouts here and there but that's what I tried to avoid when changing CI variable, and I was happy to see things work.. but not on all i386 builders :) May we close that bug for now ? F. signature.asc Description: PGP signature
Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
Hi, On Thu, 24 Dec 2020 00:05:30 +0100 Sebastian Ramacher wrote: > Source: os-autoinst > Version: 4.6.1604525166.912dfbd-0.2 > Severity: serious > Tags: ftbfs > Justification: fails to build from source (but built successfully in the past) > X-Debbugs-Cc: sramac...@debian.org > > os-autoinst FTBFS on i386: > | 3: ./29-backend-generalhw.t . ok > | 3: > | 3: # Failed test 'Result in testresults/result-reload_needles.json is ok' > | 3: # at ./99-full-stack.t line 86. > | 3: # got: 'fail' > | 3: # expected: 'ok' > | 3: Bailout called. Further testing stopped: > testresults/result-reload_needles.json failed > | 3: FAILED--Further testing stopped: testresults/result-reload_needles.json > failed > | 3/3 Test #3: test-perl-testsuite ..***Failed 332.81 sec > > See > https://buildd.debian.org/status/fetch.php?pkg=os-autoinst=i386=4.6.1604525166.912dfbd-0.2=1608714762=0 I don't know if anything changed, but the FORTH build was successful. Could be that the test is flaky. Paul OpenPGP_signature Description: OpenPGP digital signature
Bug#977990: os-autoinst: FTBFS on i386: 3/3 Test #3: test-perl-testsuite ..............***Failed 332.81 sec
Source: os-autoinst Version: 4.6.1604525166.912dfbd-0.2 Severity: serious Tags: ftbfs Justification: fails to build from source (but built successfully in the past) X-Debbugs-Cc: sramac...@debian.org os-autoinst FTBFS on i386: | 3: ./29-backend-generalhw.t . ok | 3: | 3: # Failed test 'Result in testresults/result-reload_needles.json is ok' | 3: # at ./99-full-stack.t line 86. | 3: # got: 'fail' | 3: # expected: 'ok' | 3: Bailout called. Further testing stopped: testresults/result-reload_needles.json failed | 3: FAILED--Further testing stopped: testresults/result-reload_needles.json failed | 3/3 Test #3: test-perl-testsuite ..***Failed 332.81 sec See https://buildd.debian.org/status/fetch.php?pkg=os-autoinst=i386=4.6.1604525166.912dfbd-0.2=1608714762=0 Cheers -- Sebastian Ramacher signature.asc Description: PGP signature