Discussion:
upcoming systemd-239 release — call for testing
(too old to reply)
Zbigniew Jędrzejewski-Szmek
2018-06-18 12:16:24 UTC
Permalink
Hi all,

we plan to release systemd-239 wednesday-ish and it will be landing in
rawhide. There's a bunch of new functionality, see
https://github.com/systemd/systemd/blob/master/NEWS. As always, the
majority of commits is cleanups and bugfixes and the polishing of
existing functionality. A big new feature is "portable services",
currently in preview mode. There are man pages, but an introductory
blog story is planned around the time of the release, so you might
want to wait for that.

Please give it a try and report any bugs either in bugzilla [1] or
upstream [2]. For testing, rpms are available from copr:
https://copr.fedorainfracloud.org/coprs/zbyszek/systemd/build/768345/.

Zbyszek


PS. There seems to be a regression in copr with building from src.fp.o
git. Previously I was able to point copr at the right git URL and it
would build the package. That fails [3] now with a strange error about
missing "http://None" prefix. Is this a known problem?

[1] https://bugzilla.redhat.com/enter_bug.cgi?product=Fedora&component=systemd&release=rawhide
[2] https://github.com/systemd/systemd/issues/new
[3] https://copr.fedorainfracloud.org/coprs/zbyszek/systemd/build/768342/
http://copr-dist-git.fedorainfracloud.org/per-task-logs/768342.log
https://copr-be.cloud.fedoraproject.org/results/zbyszek/systemd/srpm-builds/00768342/builder-live.log
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/OSGKZYCVS5Z3QSEVN72H3OCWHQ7FQVY
Miroslav Suchý
2018-06-18 12:54:07 UTC
Permalink
Post by Zbigniew Jędrzejewski-Szmek
PS. There seems to be a regression in copr with building from src.fp.o
git. Previously I was able to point copr at the right git URL and it
would build the package. That fails [3] now with a strange error about
missing "http://None" prefix. Is this a known problem?
Copr cannot construct SRPM because of:

Wrote: /var/lib/copr-rpmbuild/results/tmpf3uvor6m/systemd.spec
stderr: error: Unable to open /var/lib/copr-rpmbuild/results/tmpf3uvor6m/triggers.systemd: No such file or directory
can't parse specfile

triggers.systemd is not in the tarbal and not in the dist-git.

You can debug it locally by executing

/usr/bin/copr-rpmbuild --verbose --drop-resultdir --srpm --build-id 768342

on your workstation and check content of the directories you get in output.

Miroslav
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproj
Zbigniew Jędrzejewski-Szmek
2018-06-18 15:42:36 UTC
Permalink
Post by Miroslav Suchý
Post by Zbigniew Jędrzejewski-Szmek
PS. There seems to be a regression in copr with building from src.fp.o
git. Previously I was able to point copr at the right git URL and it
would build the package. That fails [3] now with a strange error about
missing "http://None" prefix. Is this a known problem?
Wrote: /var/lib/copr-rpmbuild/results/tmpf3uvor6m/systemd.spec
stderr: error: Unable to open /var/lib/copr-rpmbuild/results/tmpf3uvor6m/triggers.systemd: No such file or directory
can't parse specfile
triggers.systemd is not in the tarbal and not in the dist-git.
Thanks for looking into this.

rpm does not allow reading that file from the tarball, because it
wants it to be available for %include when the spec file is parsed.
That's why I put the in git directly.

It *is* in dist-git:
https://src.fedoraproject.org/rpms/systemd/blob/master/f/triggers.systemd

This certainly worked before with that file being there. I added that file
in April 2016, and I built systemd in copr successfully as recently as
2 months ago. This seems to be a regression.

Zbyszek
Post by Miroslav Suchý
You can debug it locally by executing
/usr/bin/copr-rpmbuild --verbose --drop-resultdir --srpm --build-id 768342
on your workstation and check content of the directories you get in output.
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/PTWNI
Jason L Tibbitts III
2018-06-18 18:34:53 UTC
Permalink
ZJ> rpm does not allow reading that file from the tarball, because it
ZJ> wants it to be available for %include when the spec file is parsed.
ZJ> That's why I put the in git directly.

This one of several reasons why using %include in a Fedora specfile is
such a terrible idea. Please just don't do it. It renders the spec
unusable unless you have a git checkout or the srpm, and thus makes a
bunch of things that you could do from, say, the nightly specfile
tarball (https://src.fedoraproject.org/repo/rpm-specs-latest.tar.xz)
rather more difficult.

- J<
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/XPU4UMEA32XJQSWWPTT2X55YPHN65BBA
Zbigniew Jędrzejewski-Szmek
2018-06-18 20:23:18 UTC
Permalink
Post by Jason L Tibbitts III
ZJ> rpm does not allow reading that file from the tarball, because it
ZJ> wants it to be available for %include when the spec file is parsed.
ZJ> That's why I put the in git directly.
This one of several reasons why using %include in a Fedora specfile is
such a terrible idea. Please just don't do it. It renders the spec
unusable unless you have a git checkout or the srpm, and thus makes a
bunch of things that you could do from, say, the nightly specfile
tarball (https://src.fedoraproject.org/repo/rpm-specs-latest.tar.xz)
rather more difficult.
It's either %include or a fairly big copy&paste. The first solution is
awkward, and the second one is ... awkward. But yeah, maybe we should
bite the bullet and do the copy&paste. A bit sad.

Zbyszek
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/SA6HF2ZI4VVZJ7X766TCZLM2
Jason L Tibbitts III
2018-06-18 20:28:17 UTC
Permalink
ZJ> It's either %include or a fairly big copy&paste. The first solution
ZJ> is awkward, and the second one is ... awkward. But yeah, maybe we
ZJ> should bite the bullet and do the copy&paste. A bit sad.

It shouldn't be that bad. The body of the scriptlet shouldn't be large.
If it is, then put it in its own, install that, and have the scriptlet
call it.

I went to look at why the scriptlets are so large, but of course I'm
looking at the spec tarball and so all I can see is %include %{SOURCE1}.
Oh, well.

- J<
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/U33JU36MBW5Q5PSHA
Zbigniew Jędrzejewski-Szmek
2018-06-18 20:35:06 UTC
Permalink
Post by Jason L Tibbitts III
ZJ> It's either %include or a fairly big copy&paste. The first solution
ZJ> is awkward, and the second one is ... awkward. But yeah, maybe we
ZJ> should bite the bullet and do the copy&paste. A bit sad.
It shouldn't be that bad. The body of the scriptlet shouldn't be large.
If it is, then put it in its own, install that, and have the scriptlet
call it.
That's not an attractive option, because there's so many of them.
Also, the scriptlets are (were, it's a longer story) in lua, executed
inline. Moving that out to a separate file would break that.
See https://github.com/systemd/systemd/blob/master/src/core/triggers.systemd.in.
Post by Jason L Tibbitts III
I went to look at why the scriptlets are so large, but of course I'm
looking at the spec tarball and so all I can see is %include %{SOURCE1}.
Oh, well.
Zbyszek
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedo
Jason L Tibbitts III
2018-06-18 20:51:19 UTC
Permalink
ZJ> That's not an attractive option, because there's so many of them.

Ah, so the issue is merely the quantity of scripts. I did check out the
package from git and it doesn't seem all that bad; most of what's there
is (sort of excessively verbose) comments and without them you just have
54 lines. In lua it would be a bit more verbose, certainly, though
interestingly most of what's there is boilerplate and makes me wonder if
most of that could be generated by a macro.

Unfortunately rpm just isn't optimized for the number of things the
systemd package needs to do. Any solution is probably going to be a bit
awkward.

- J<
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/E3UUGJ5T3WAQBG3
Adam Williamson
2018-06-20 05:22:27 UTC
Permalink
Post by Zbigniew Jędrzejewski-Szmek
Hi all,
we plan to release systemd-239 wednesday-ish and it will be landing in
rawhide. There's a bunch of new functionality, see
https://github.com/systemd/systemd/blob/master/NEWS. As always, the
majority of commits is cleanups and bugfixes and the polishing of
existing functionality. A big new feature is "portable services",
currently in preview mode. There are man pages, but an introductory
blog story is planned around the time of the release, so you might
want to wait for that.
Please give it a try and report any bugs either in bugzilla [1] or
https://copr.fedorainfracloud.org/coprs/zbyszek/systemd/build/768345/.
Thanks for the heads up on this. I built a test Rawhide netinst image
with the updated systemd and ran it through openQA. It mostly worked,
but several of the tests failed with this error in anaconda:

22:13:42,713 INF threading: Thread Failed: AnaInstallThread (140332673541888)
22:13:42,713 DBG exception: running handleException
22:13:42,716 CRT exception: Traceback (most recent call last):

File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 843, in wrapped
ret = orig_obj(*args, **kwargs)

File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 465, in lvm_vgcreate
return _lvm_vgcreate(name, pv_list, pe_size, extra)

GLib.GError: g-dbus-error-quark: Waiting for 'VgCreate' method of the '/com/redhat/lvmdbus1/Manager' object to finish failed: Failed to get Complete property of the / object: GDBus.Error:org.freedesktop.DBus.Error.UnknownMethod: Method "Get" with signature "ss" on interface "org.freedesktop.DBus.Properties" doesn't exist
(19)

it seems like an intermittent error, as not all the tests that create
LVM VGs failed, and restarting the failed tests mostly resulted in
passes...but I don't think we've seen this particular error, even as an
intermittent one, in recent Rawhide tests with systemd 238, so it seems
like it *could* be caused by the update. Any idea where this may be
coming from?

Here's one of the failed tests:

https://openqa.stg.fedoraproject.org/tests/318564

all log files etc. can be found on the "Logs & Assets" tab.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/4EQTD
Martin Kolman
2018-06-20 09:40:17 UTC
Permalink
Post by Adam Williamson
Post by Zbigniew Jędrzejewski-Szmek
Hi all,
we plan to release systemd-239 wednesday-ish and it will be landing in
rawhide. There's a bunch of new functionality, see
https://github.com/systemd/systemd/blob/master/NEWS. As always, the
majority of commits is cleanups and bugfixes and the polishing of
existing functionality. A big new feature is "portable services",
currently in preview mode. There are man pages, but an introductory
blog story is planned around the time of the release, so you might
want to wait for that.
Please give it a try and report any bugs either in bugzilla [1] or
https://copr.fedorainfracloud.org/coprs/zbyszek/systemd/build/768345/.
Thanks for the heads up on this. I built a test Rawhide netinst image
with the updated systemd and ran it through openQA. It mostly worked,
22:13:42,713 INF threading: Thread Failed: AnaInstallThread (140332673541888)
22:13:42,713 DBG exception: running handleException
File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 843, in wrapped
ret = orig_obj(*args, **kwargs)
File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 465, in lvm_vgcreate
return _lvm_vgcreate(name, pv_list, pe_size, extra)
GLib.GError: g-dbus-error-quark: Waiting for 'VgCreate' method of the '/com/redhat/lvmdbus1/Manager' object to finish
failed: Failed to get Complete property of the / object: GDBus.Error:org.freedesktop.DBus.Error.UnknownMethod: Method
"Get" with signature "ss" on interface "org.freedesktop.DBus.Properties" doesn't exist
(19)
it seems like an intermittent error, as not all the tests that create
LVM VGs failed, and restarting the failed tests mostly resulted in
passes...but I don't think we've seen this particular error, even as an
intermittent one, in recent Rawhide tests with systemd 238, so it seems
like it *could* be caused by the update. Any idea where this may be
coming from?
Adding Vojta and Dave to CC to take a look at this.
Post by Adam Williamson
https://openqa.stg.fedoraproject.org/tests/318564
all log files etc. can be found on the "Logs & Assets" tab.
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/U6NTJ
Lennart Poettering
2018-06-20 11:17:45 UTC
Permalink
Post by Adam Williamson
Post by Zbigniew Jędrzejewski-Szmek
Hi all,
we plan to release systemd-239 wednesday-ish and it will be landing in
rawhide. There's a bunch of new functionality, see
https://github.com/systemd/systemd/blob/master/NEWS. As always, the
majority of commits is cleanups and bugfixes and the polishing of
existing functionality. A big new feature is "portable services",
currently in preview mode. There are man pages, but an introductory
blog story is planned around the time of the release, so you might
want to wait for that.
Please give it a try and report any bugs either in bugzilla [1] or
https://copr.fedorainfracloud.org/coprs/zbyszek/systemd/build/768345/.
Thanks for the heads up on this. I built a test Rawhide netinst image
with the updated systemd and ran it through openQA. It mostly worked,
22:13:42,713 INF threading: Thread Failed: AnaInstallThread (140332673541888)
22:13:42,713 DBG exception: running handleException
File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 843, in wrapped
ret = orig_obj(*args, **kwargs)
File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 465, in lvm_vgcreate
return _lvm_vgcreate(name, pv_list, pe_size, extra)
GLib.GError: g-dbus-error-quark: Waiting for 'VgCreate' method of the '/com/redhat/lvmdbus1/Manager' object to finish failed: Failed to get Complete property of the / object: GDBus.Error:org.freedesktop.DBus.Error.UnknownMethod: Method "Get" with signature "ss" on interface "org.freedesktop.DBus.Properties" doesn't exist
(19)
it seems like an intermittent error, as not all the tests that create
LVM VGs failed, and restarting the failed tests mostly resulted in
passes...but I don't think we've seen this particular error, even as an
intermittent one, in recent Rawhide tests with systemd 238, so it seems
like it *could* be caused by the update. Any idea where this may be
coming from?
Hmm, this doesn't look like a systemd issue to me. This is a dbus
method call failure of some form. The dbus client and the dbus service
in this case are not provided by systemd, and neither is the dbus
message bus in between. Hence systemd doesn't really insert itself in
any form in the ongoing communication between this IPC client and
server.

That said, systemd is involved in service activation if the service
isn't around yet at the time the client asks for the service. However,
judging by the provided logs
(https://openqa.stg.fedoraproject.org/tests/318564/file/_do_install_and_reboot-syslog)
the activation of the lvmdbus1 service actually completes fine.

Hence, this is somewhere between the message bus, and the LVM client
and server.

Lennart

--
Lennart Poettering, Red Hat
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/QL7GBBIIF3C3E72G273DQHGF
Adam Williamson
2018-06-20 23:56:23 UTC
Permalink
Post by Lennart Poettering
Post by Adam Williamson
Post by Zbigniew Jędrzejewski-Szmek
Hi all,
we plan to release systemd-239 wednesday-ish and it will be landing in
rawhide. There's a bunch of new functionality, see
https://github.com/systemd/systemd/blob/master/NEWS. As always, the
majority of commits is cleanups and bugfixes and the polishing of
existing functionality. A big new feature is "portable services",
currently in preview mode. There are man pages, but an introductory
blog story is planned around the time of the release, so you might
want to wait for that.
Please give it a try and report any bugs either in bugzilla [1] or
https://copr.fedorainfracloud.org/coprs/zbyszek/systemd/build/768345/.
Thanks for the heads up on this. I built a test Rawhide netinst image
with the updated systemd and ran it through openQA. It mostly worked,
22:13:42,713 INF threading: Thread Failed: AnaInstallThread (140332673541888)
22:13:42,713 DBG exception: running handleException
File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 843, in wrapped
ret = orig_obj(*args, **kwargs)
File "/usr/lib64/python3.6/site-packages/gi/overrides/BlockDev.py", line 465, in lvm_vgcreate
return _lvm_vgcreate(name, pv_list, pe_size, extra)
GLib.GError: g-dbus-error-quark: Waiting for 'VgCreate' method of the '/com/redhat/lvmdbus1/Manager' object to finish failed: Failed to get Complete property of the / object: GDBus.Error:org.freedesktop.DBus.Error.UnknownMethod: Method "Get" with signature "ss" on interface "org.freedesktop.DBus.Properties" doesn't exist
(19)
it seems like an intermittent error, as not all the tests that create
LVM VGs failed, and restarting the failed tests mostly resulted in
passes...but I don't think we've seen this particular error, even as an
intermittent one, in recent Rawhide tests with systemd 238, so it seems
like it *could* be caused by the update. Any idea where this may be
coming from?
Hmm, this doesn't look like a systemd issue to me. This is a dbus
method call failure of some form. The dbus client and the dbus service
in this case are not provided by systemd, and neither is the dbus
message bus in between. Hence systemd doesn't really insert itself in
any form in the ongoing communication between this IPC client and
server.
That said, systemd is involved in service activation if the service
isn't around yet at the time the client asks for the service. However,
judging by the provided logs
(https://openqa.stg.fedoraproject.org/tests/318564/file/_do_install_and_reboot-syslog)
the activation of the lvmdbus1 service actually completes fine.
Hence, this is somewhere between the message bus, and the LVM client
and server.
Thanks...so to try and confirm I did a run of the same set of tests
(the 'universal' tests) with a regular Rawhide netinst image (one with
systemd 238), and indeed the same bug occurred at least once:

https://openqa.stg.fedoraproject.org/tests/318862

so it does indeed not appear to be to do with systemd 239. Wonder what
is causing it, though. Hum.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net
_______________________________________________
devel mailing list -- ***@lists.fedoraproject.org
To unsubscribe send an email to devel-***@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/***@lists.fedoraproject.org/message/EQTG7EV
Loading...