Re: Feedback on pakfire-3 testing

[Adolf Belka <adolf.belka@ipfire.org>]

[-- Attachment #1: Type: text/plain, Size: 10447 bytes --]

Hi Michael,

On 10/01/2025 15:00, Michael Tremer wrote:
> Hello Adolf,
> 
> Thank you for giving this a test.
> 
>> On 10 Jan 2025, at 13:51, Adolf Belka <adolf.belka(a)ipfire.org> wrote:
>>
>> Hi Michael,
>>
>> I did a fresh git clone of the pakfire repo and built it.
>>
>> Then ran builds of beep, libxcrypt and openssl. All completed successfully.
> 
> That is good news.
> 
>> beep took 15 secs, libxcrypt 50 secs and openssl 13 mins.
> 
> Cool. This sounds about right. The first build should have created a snapshot may so that running the second would have started much faster?
> 
>> In the autogen.sh part of the pakfire build I did see a couple of warnings but it didn't stop the pakfire build from completing successfully.
>>
>>
>> You should update your 'aclocal.m4' by running aclocal.
>> configure.ac:191: warning: AC_PROG_LEX without either yywrap or noyywrap is obsolete
>>
>> parallel-tests: installing 'build-aux/test-driver'
>> Makefile.am:1175: warning: variable 'TESTSUITE_LDFLAGS' is defined but no program or
>> Makefile.am:1175: library has 'TESTSUITE' as canonical name (possible typo)
> 
> Yes, the first one I have no idea how to get rid of, the second is probably just renaming the variable.
> 
>> 8 from 30 tests in the make check section failed.
> 
> That is not right. Could you check the *.log files of the individual ones and see what is going wrong?

Here is the fail message for each of the 8 - does make check need to be run as root? I haven't been, only the make install.

main
pakfire: pakfire_create: Could not mount tmpfs: Operation not permitted
pakfire: pakfire_free: Destroying
TESTS: ERROR: Failed assertion: pakfire_create(&pakfire, t->ctx, NULL, NULL, NULL, 0) == -EINVAL tests/libpakfire/main.c:41 test_fail

build
pakfire: pakfire_create: Could not mount tmpfs: Operation not permitted
pakfire: pakfire_free: Destroying
TESTS: ERROR: Failed assertion: pakfire_build_create(&build, t->ctx, t->config, NULL, NULL, 0) (r = -1 - Operation not permitted, errno = Operation not permitted) at tests/libpakfire/build.c:31 test_create_and_free

cgroup
pakfire: pakfire_cgroup_open: Created cgroup (root)
pakfire: __pakfire_cgroup_open: Could not create cgroup 'pakfire-test': Permission denied
TESTS: ERROR: Failed assertion: pakfire_cgroup_create(&cgroup, t->ctx, NULL, "pakfire-test", 0) (r = -13 - Permission denied, errno = Permission denied) at tests/libpakfire/cgroup.c:34 test_create_and_destroy

jail
pakfire: pakfire_jail_setup_uid_mapping: Could not map UIDs: Invalid argument
TESTS: ERROR: Failed assertion: pakfire_jail_exec_command(jail, argv, NULL, 0) == 123 tests/libpakfire/jail.c:71 test_exit_code
pakfire: pakfire_jail_free: Freeing jail at 0x590fa3a18c50

makefile
pakfire: pakfire_parser_create_package: Building package from namespace '(null)'
pakfire: pakfire_parser_create_package: EVR is empty
pakfire: pakfire_parser_create_package: Could not create package: No such file or directory
TESTS: ERROR: Failed assertion: r == 0 tests/libpakfire/makefile.c:140 test_packages

os
TESTS: running test_distro
TESTS: ERROR: Failed assertion: *distro.version || *distro.version_id tests/libpakfire/os.c:109 test_distro

util
pakfire: pakfire_create:     vendor     = Test Project
pakfire: pakfire_create:     slogan     = Always code as if the guy who ends up maintaining your code will be a violent psychopath who knows where you live
pakfire: pakfire_rlimit_set: RLIMIT_NOFILE set to 524288
TESTS: ERROR: Failed assertion: pakfire_mkparentdir("/test/1/file.txt", 0755) (r = -1 - Operation not permitted, errno = Permission denied) at tests/libpakfire/util.c:43 test_mkdir

package.py
DEBUG:pakfire:    vendor     = Test Project
DEBUG:pakfire:    slogan     = Always code as if the guy who ends up maintaining your code will be a violent psychopath who knows where you live
DEBUG:pakfire:RLIMIT_NOFILE set to 524288
F
======================================================================
FAIL: test_open (__main__.PackageTest.test_open)
----------------------------------------------------------------------
Traceback (most recent call last):
   File "/home/source/pakfire/./tests/python/package.py", line 39, in test_open
     self.assertEqual(p.groups, "Applications/System")
     ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: [] != 'Applications/System'

----------------------------------------------------------------------
Ran 1 test in 0.003s

> 
>> In the package build stage for all three packages that I tested I got the following:-
>>
>> sudo pakfire-builder build beep/beep.nm
>>   Scanning local 00:00 60/60
>> Could not retrieve uname for 1000: No such file or directory
>> Could not retrieve gname for 1000: No such file or directory
>> Could not retrieve uname for 1000: No such file or directory
>> Could not retrieve gname for 1000: No such file or directory
>> Could not retrieve uname for 1000: No such file or directory
>> Could not retrieve gname for 1000: No such file or directory
> 
> Hmm, this is not right… As the files checked out by another user with ID 1000?

I forgot to mention. uid and gid 1000 is my user that is logged in running the pakfire-builder process. It is just my normal user, nothing special.

> 
>>   beep-1.4.12.tar.gz 87k     87kB/s 00:01   0%
>>   beep-1.4.12-1.ipfire3.src 88k      0B/s 100%
>> Building beep-1.4.12-1.ipfire3.src (eed8a645-191c-46d4-91a5-dfc82aec9829)...
>>
>>
>>   python3-gobject3-3.46.0-1.ipfire3.x86_64 100%
>> #  This file is part of systemd.
>> #
>> #  systemd is free software; you can redistribute it and/or modify it
>> #  under the terms of the GNU Lesser General Public License as published by
>> #  the Free Software Foundation; either version 2.1 of the License, or
>> #  (at your option) any later version.
>>
>> # The superuser
>> g root    0       -            -
>> u root    0:0     "Super User" /root
>>
>> # The nobody user/group for NFS file systems
>> g nobody 65534       -            -
>> u nobody 65534:65534 "Kernel Overflow User"     -
>>
>> # Administrator group: can *see* more than normal users
>> g adm     4     -            -
>>
>> # Administrator group: can *do* more than normal users
>> g wheel   10     -            -
>>
>> # Access to shared database of users on the system
>> g utmp    22     -            -
>>
>> # Physical and virtual hardware access groups
>> g audio   63     -            -
>> g cdrom   11     -            -
>> g dialout 18     -            -
>> g disk    6     -            -
>> g input   104     -            -
>> g kmem    9     -            -
>> g kvm     36     -            -
>> g lp      7     -            -
>> g render  105     -            -
>> g sgx     106     -            -
>> g tape    33     -            -
>> g tty     5     -            -
>> g video   39     -            -
>>
>> # Default group for normal users
>> g users   100     -            -
>> #  This file is part of systemd.
>> #
>> #  systemd is free software; you can redistribute it and/or modify it
>> #  under the terms of the GNU Lesser General Public License as published by
>> #  the Free Software Foundation; either version 2.1 of the License, or
>> #  (at your option) any later version.
>>
>> u systemd-coredump - "systemd Core Dumper"
>> #  This file is part of systemd.
>> #
>> #  systemd is free software; you can redistribute it and/or modify it
>> #  under the terms of the GNU Lesser General Public License as published by
>> #  the Free Software Foundation; either version 2.1 of the License, or
>> #  (at your option) any later version.
>>
>> g systemd-journal 190 -
>> #  This file is part of systemd.
>> #
>> #  systemd is free software; you can redistribute it and/or modify it
>> #  under the terms of the GNU Lesser General Public License as published by
>> #  the Free Software Foundation; either version 2.1 of the License, or
>> #  (at your option) any later version.
>>
>> u systemd-oom - "systemd Userspace OOM Killer"
> 
> This is from a systemd feature that currently seems to be very broken, I will put this on my TODO list.
> 
>>   systemd-254-3.ipfire3.x86_64 100%
>> *** buffer overflow detected ***: terminated
> 
> This is also a problem inside the build environment and I believe rebuilding systemd fixes it.
> 
>> The child process terminated abnormally with status code 11
>>   dbus-broker-32-1.ipfire3.x86_64 100%
>>   uuidd-2.39.2-3.ipfire3.x86_64 100%
>>   setup-3.0-19.ipfire3.noarch 100%
>> *** buffer overflow detected ***: terminated
>> The child process terminated abnormally with status code 11
>>   sudo-1.9.14p3-2.ipfire3.x86_64 100%
>> *** buffer overflow detected ***: terminated
>> The child process terminated abnormally with status code 11
>>   bash-completion-2.11-1.ipfire3.noarch 100%
>>   filesystem-002-5.ipfire3.x86_64 0%
>>   groff-1.22.4-1.ipfire3.x86_64 100%
>>   openssh-9.4p1-1.ipfire3.x86_64 100%
>>   smartmontools-7.4-1.ipfire3.x86_64 100%
>>   mc-4.8.29-3.ipfire3.x86_64 100%
>>   libtool-2.4.7-2.ipfire3.x86_64 100%
>>   gettext-0.21.1-5.1.ipfire3.x86_64 100%
>>   gettext-libs-0.21.1-5.1.ipfire3.x86_64 100%
>>   findutils-4.9.0-1.ipfire3.x86_64 100%
>>   gcc-12.3.0-1.ipfire3.x86_64 100%
>>   gcc-c++-12.3.0-1.ipfire3.x86_64 100%
>>   cpp-12.3.0-1.ipfire3.x86_64 100%
>>   man-db-2.11.2-1.ipfire3.x86_64 100%
>> *** buffer overflow detected ***: terminated
>> The child process terminated abnormally with status code 11
>>   ipfire-release-build-3.0-alpha2-10.ipfire3.noarch 0%
>>   pakfire-0.9.29-1.ipfire3.x86_64 100%
>>   ipfire-release-3.0-alpha2-10.ipfire3.noarch 100%
>> *** buffer overflow detected ***: terminated
>> The child process terminated abnormally with status code 11
>> Running in chroot, ignoring command 'start'
>> Running build stage 'prepare'
>>
>> I recall you saying that running a repeat of the build should result in a quicker build. I have not found that. It has stayed very similar in time in all three cases.
> 
> Can you run “pakfire-build snapshot update” and see if that creates the snapshot? Maybe I didn’t fully think this through when it is being created.

That made the improvement. Now beep built in 2 seconds and libxcrypt in 35 seconds.

Regards,
Adolf.
> 
> -Michael
> 
>>
>> Regards,
>>
>> Adolf.
>>
>