Bug#864297: unblock: schroot/1.6.10-4

[Adam D. Barratt]

On Wed, 2017-10-04 at 14:24 +0200, Cyril Brulebois wrote:
> Raphael Hertzog  (2017-10-04):
> > Thanks, uploaded (as a source-only upload, I hope it's fine with
> > spu
> > too).
> 
> AFAICT jessie doesn't support source-only, anything newer does.
> 

That's correct. stretch is the first stable release that supports
source-only uploads.

Regards,

Adam

Bug#864297: unblock: schroot/1.6.10-4

[Adam D. Barratt]

Control: tags -1 + pending

On Wed, 2017-10-04 at 10:21 +0200, Raphael Hertzog wrote:
> Hi,
> 
> On Fri, 29 Sep 2017, Adam D. Barratt wrote:
> > > We already missed the first point release and the change has now
> > > been
> > > in unstable for a long time already.
> > 
> > Mostly on the basis that the changes have persisted "as is" in
> > unstable, I guess doing so is probably better than nothing. Please
> > go
> > ahead.
> 
> Thanks, uploaded

Flagged for acceptance.

Regards,

Adam

Bug#864297: unblock: schroot/1.6.10-4

[Cyril Brulebois]

Raphael Hertzog  (2017-10-04):
> Thanks, uploaded (as a source-only upload, I hope it's fine with spu
> too).

AFAICT jessie doesn't support source-only, anything newer does.


KiBi.


signature.asc
Description: PGP signature

Bug#864297: unblock: schroot/1.6.10-4

[Raphael Hertzog]

Hi,

On Fri, 29 Sep 2017, Adam D. Barratt wrote:
> > We already missed the first point release and the change has now been
> > in unstable for a long time already.
> 
> Mostly on the basis that the changes have persisted "as is" in
> unstable, I guess doing so is probably better than nothing. Please go
> ahead.

Thanks, uploaded (as a source-only upload, I hope it's fine with spu too).

Cheers,
-- 
Raphaël Hertzog ◈ Debian Developer

Support Debian LTS: https://www.freexian.com/services/debian-lts.html
Learn to master Debian: https://debian-handbook.info/get/

Bug#864297: unblock: schroot/1.6.10-4

[Adam D. Barratt]

Control: tags -1 + confirmed

On Thu, 2017-09-14 at 11:53 +0200, Raphael Hertzog wrote:
> Hello,
> 
> On Fri, 30 Jun 2017, Cyril Brulebois wrote:
> > > So let me know what I should upload:
> > > - only the bash-completion fix
> > > - or the whole update that I wanted to push
> > 
> > Michael, any comments on the proposed systemd change for jessie,
> > please?
> 
> Michael said that he had nothing to add. Can someone please let me go
> forward with this proposed update ?

Well, his comments didn't seem all that positive. :-(

> We already missed the first point release and the change has now been
> in unstable for a long time already.
> 

Mostly on the basis that the changes have persisted "as is" in
unstable, I guess doing so is probably better than nothing. Please go
ahead.

Regards,

Adam

Bug#864297: unblock: schroot/1.6.10-4

[Raphael Hertzog]

Hello,

On Fri, 30 Jun 2017, Cyril Brulebois wrote:
> > So let me know what I should upload:
> > - only the bash-completion fix
> > - or the whole update that I wanted to push
> 
> Michael, any comments on the proposed systemd change for jessie, please?

Michael said that he had nothing to add. Can someone please let me go
forward with this proposed update ?

We already missed the first point release and the change has now been in
unstable for a long time already.

Cheers,
-- 
Raphaël Hertzog ◈ Debian Developer

Support Debian LTS: https://www.freexian.com/services/debian-lts.html
Learn to master Debian: https://debian-handbook.info/get/

Bug#864297: unblock: schroot/1.6.10-4

[Cyril Brulebois]

Hi,

Raphael Hertzog  (2017-06-29):
> On Tue, 27 Jun 2017, Cyril Brulebois wrote:
> > Personally, I find the systemd part a bit touchy, but I'm happy to ACK
> > the bash-completion fix on its own. Maybe we could handle this in two
> > steps, until the systemd unit situation settles (#865110)?
> 
> I consider #865110 to be a minor cleanup that is rather low on my
> TODO list.
> 
> If the systemd part has to wait until we have fixed #865110 in unstable
> then I'm afraid that I will not bother doing a second stable upload just
> for this.
> 
> So let me know what I should upload:
> - only the bash-completion fix
> - or the whole update that I wanted to push

Michael, any comments on the proposed systemd change for jessie, please?


KiBi.


signature.asc
Description: Digital signature

Bug#864297: unblock: schroot/1.6.10-4

[Raphael Hertzog]

Hi,

On Tue, 27 Jun 2017, Cyril Brulebois wrote:
> Personally, I find the systemd part a bit touchy, but I'm happy to ACK
> the bash-completion fix on its own. Maybe we could handle this in two
> steps, until the systemd unit situation settles (#865110)?

I consider #865110 to be a minor cleanup that is rather low on my
TODO list.

If the systemd part has to wait until we have fixed #865110 in unstable
then I'm afraid that I will not bother doing a second stable upload just
for this.

So let me know what I should upload:
- only the bash-completion fix
- or the whole update that I wanted to push

Thank you!
-- 
Raphaël Hertzog ◈ Debian Developer

Support Debian LTS: https://www.freexian.com/services/debian-lts.html
Learn to master Debian: https://debian-handbook.info/get/

Bug#864297: unblock: schroot/1.6.10-4

[Cyril Brulebois]

Raphael Hertzog  (2017-06-19):
> On Mon, 19 Jun 2017, Michael Biebl wrote:
> > Given that, I wonder if a stable upload is actually still needed.
> 
> The stable upload also fixes the bash completion file and fixes the
> dependencies of the service file to not depend on the systemd version
> of $network in the LSB dependencies.

Personally, I find the systemd part a bit touchy, but I'm happy to ACK
the bash-completion fix on its own. Maybe we could handle this in two
steps, until the systemd unit situation settles (#865110)?


KiBi.


signature.asc
Description: Digital signature

Bug#835104: Bug#864297: unblock: schroot/1.6.10-4

[Aurelien Jarno]

On 2017-06-19 15:08, Raphael Hertzog wrote:
> [ Putting Aurélion in copy since he's the one who requested this service
> file ]
> 
> On Sun, 18 Jun 2017, Michael Biebl wrote:
> > Am 17.06.2017 um 22:27 schrieb Michael Biebl:
> > 
> > > For that create a file
> > > /lib/systemd/system/schroot.service.d/timeout.conf (feel free to choose
> > > a better name), containing:
> > > 
> > > [Service]
> > > Type=oneshot
> > > TimeoutSec=infinity
> > 
> > If you only need to change the timeout, I probably wouldn't change Type=
> > at all, but only set TimeoutSec= via a drop-in.
> > The default is 5min for SysV init scripts.
> > I'm surprised that #835104 says that the init scripts are killed after
> > 90sec though.
> > Can schroot take longer then 5min to start or stop? Do we really need to
> > disable the timeout completely (infinity) or would simply increasing the
> > value to say 15min be sufficient?
> 
> Aurélien, can you answer Michael's question ?

DSA uses this init script with SESSIONS_RECOVER="end" in
/etc/default/schroot. When there are dozen of chroots opened (which is
common on the porterboxes), the current script takes more than 5 minutes
to cleanup all the chroots upon reboot. I guess increasing the timeout is
fine, 15min would already be a good start.

Aurelien

-- 
Aurelien Jarno  GPG: 4096R/1DDD8C9B
aurel...@aurel32.net http://www.aurel32.net

Bug#864297: unblock: schroot/1.6.10-4

[Raphael Hertzog]

On Mon, 19 Jun 2017, Michael Biebl wrote:
> Given that, I wonder if a stable upload is actually still needed.

The stable upload also fixes the bash completion file and fixes
the dependencies of the service file to not depend on the systemd version
of $network in the LSB dependencies.

Cheers,
-- 
Raphaël Hertzog ◈ Debian Developer

Support Debian LTS: https://www.freexian.com/services/debian-lts.html
Learn to master Debian: https://debian-handbook.info/get/

Bug#835104: Bug#864297: unblock: schroot/1.6.10-4

[Raphael Hertzog]

[ Putting Aurélion in copy since he's the one who requested this service
file ]

On Sun, 18 Jun 2017, Michael Biebl wrote:
> Am 17.06.2017 um 22:27 schrieb Michael Biebl:
> 
> > For that create a file
> > /lib/systemd/system/schroot.service.d/timeout.conf (feel free to choose
> > a better name), containing:
> > 
> > [Service]
> > Type=oneshot
> > TimeoutSec=infinity
> 
> If you only need to change the timeout, I probably wouldn't change Type=
> at all, but only set TimeoutSec= via a drop-in.
> The default is 5min for SysV init scripts.
> I'm surprised that #835104 says that the init scripts are killed after
> 90sec though.
> Can schroot take longer then 5min to start or stop? Do we really need to
> disable the timeout completely (infinity) or would simply increasing the
> value to say 15min be sufficient?

Aurélien, can you answer Michael's question ?

Cheers,
-- 
Raphaël Hertzog ◈ Debian Developer

Support Debian LTS: https://www.freexian.com/services/debian-lts.html
Learn to master Debian: https://debian-handbook.info/get/

Bug#864297: unblock: schroot/1.6.10-4

[Michael Biebl]

Am 19.06.2017 um 14:11 schrieb Raphael Hertzog:
> Hello Michael,
> 
> On Sat, 17 Jun 2017, Michael Biebl wrote:
>> I would suggest to simply ship a drop-in snippet for schroot, which sets
>> the Type and Timeout.
> 
> Thanks for the suggestion but I'll skip it. This is what I would have
> expected DSA to do if I had not shipped a systemd service unit. 

Maybe the DSA admins weren't aware that they could simply ship such a
drop-in easily to increase the timeout. Should be trivial to distribute
that via puppet.
Given that, I wonder if a stable upload is actually still needed.

>> It also makes it more obvious that eventually this package should be
>> converted to use proper service files. I always cringe at packages
>> shipping "native" service files which only run the SysV init scripts.
> 
> Feel free to open a separate bug report if you want to make sure that we
> don't forget this.

Done, #865110


-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature

Bug#864297: unblock: schroot/1.6.10-4

[Raphael Hertzog]

Hello Michael,

On Sat, 17 Jun 2017, Michael Biebl wrote:
> I would suggest to simply ship a drop-in snippet for schroot, which sets
> the Type and Timeout.

Thanks for the suggestion but I'll skip it. This is what I would have
expected DSA to do if I had not shipped a systemd service unit. But now
I have a full unit in unstable already and I would have to change first in
unstable before being able to backport the change in stretch.

> It also makes it more obvious that eventually this package should be
> converted to use proper service files. I always cringe at packages
> shipping "native" service files which only run the SysV init scripts.

Feel free to open a separate bug report if you want to make sure that we
don't forget this.

Cheers,
-- 
Raphaël Hertzog ◈ Debian Developer

Support Debian LTS: https://www.freexian.com/services/debian-lts.html
Learn to master Debian: https://debian-handbook.info/get/

Bug#835104: Bug#864297: unblock: schroot/1.6.10-4

[Michael Biebl]

Am 17.06.2017 um 22:27 schrieb Michael Biebl:

> For that create a file
> /lib/systemd/system/schroot.service.d/timeout.conf (feel free to choose
> a better name), containing:
> 
> [Service]
> Type=oneshot
> TimeoutSec=infinity

If you only need to change the timeout, I probably wouldn't change Type=
at all, but only set TimeoutSec= via a drop-in.
The default is 5min for SysV init scripts.
I'm surprised that #835104 says that the init scripts are killed after
90sec though.
Can schroot take longer then 5min to start or stop? Do we really need to
disable the timeout completely (infinity) or would simply increasing the
value to say 15min be sufficient?


Michael

-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature

Bug#864297: unblock: schroot/1.6.10-4

[Michael Biebl]

Hi Raphaël,

Am 17.06.2017 um 19:59 schrieb Michael Biebl:
> Am 06.06.2017 um 17:31 schrieb Raphaël Hertzog:
>> +[Service]
>> +Type=oneshot
...
> Hm, so you just wrapped the sysv init script, but didn't actually
> provide a native systemd unit.
> Why is the unit generated by systemd-sysv-generator not sufficient?

Julien pointed out on IRC, that you changed to Type=oneshot to disable
the start/stop timeouts and probably because Type=oneshot is more
appropriate for schroot. The reasoning for the latter should be
documented in the changelog though.

I would suggest to simply ship a drop-in snippet for schroot, which sets
the Type and Timeout.

For that create a file
/lib/systemd/system/schroot.service.d/timeout.conf (feel free to choose
a better name), containing:

[Service]
Type=oneshot
TimeoutSec=infinity


Unfortunately, you currently need to explicitly specifiy the timeout for
SysV generated service files due to [1], i.e. simply setting
Type=oneshot for SysV generated unit files is not sufficient.

I thinks this would be a nicer solution as it is more minimal and
focuses on the actual necessary changes.
It also makes it more obvious that eventually this package should be
converted to use proper service files. I always cringe at packages
shipping "native" service files which only run the SysV init scripts.

Regards,
Michael


[1] https://github.com/systemd/systemd/issues/6143

-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature

Bug#864297: unblock: schroot/1.6.10-4

[Michael Biebl]

Am 06.06.2017 um 17:31 schrieb Raphaël Hertzog:

>  .PHONY: override_dh_auto_configure override_dh_auto_clean 
> override_dh_auto_build override_dh_auto_test override_dh_auto_install 
> override_dh_installchangelogs override_dh_installinit install-arch 
> install-indep
> diff -Nru schroot-1.6.10/debian/schroot.service 
> schroot-1.6.10/debian/schroot.service
> --- schroot-1.6.10/debian/schroot.service 1970-01-01 01:00:00.0 
> +0100
> +++ schroot-1.6.10/debian/schroot.service 2017-06-06 16:38:31.0 
> +0200
> @@ -0,0 +1,13 @@
> +[Unit]
> +Description=Recover schroot sessions
> +Documentation=man:schroot(1)
> +RequiresMountsFor=/var/lib/schroot
> +
> +[Service]
> +Type=oneshot
> +ExecStart=/etc/init.d/schroot start
> +ExecStop=/etc/init.d/schroot stop
> +RemainAfterExit=yes

Hm, so you just wrapped the sysv init script, but didn't actually
provide a native systemd unit.
Why is the unit generated by systemd-sysv-generator not sufficient?


-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?



signature.asc
Description: OpenPGP digital signature

Bug#864297: unblock: schroot/1.6.10-4

[Jonathan Wiltshire]

On Tue, Jun 06, 2017 at 05:31:36PM +0200, Raphaël Hertzog wrote:
> It might not qualify for last minutes updates and I would understand
> if you ask me to handle this via stretch-pu in the future but I would like
> to have two bugs fixed in stretch:
> 
> - the completion code is completely broken (#855283)
> - a member of the DSA team asked me to provide a systemd unit to avoid
>   timeouts they encounter on porter boxes (#835104)

I would rather that both these saw more testing in unstable first, so
please work with SRMs for a later stable update.

Thanks,

-- 
Jonathan Wiltshire  j...@debian.org
Debian Developer http://people.debian.org/~jmw

4096R: 0xD3524C51 / 0A55 B7C5 1223 3942 86EC  74C3 5394 479D D352 4C51

Bug#864297: unblock: schroot/1.6.10-4

[Raphaël Hertzog]

Package: release.debian.org
Severity: normal
User: release.debian@packages.debian.org
Usertags: unblock

Please unblock package schroot

It might not qualify for last minutes updates and I would understand
if you ask me to handle this via stretch-pu in the future but I would like
to have two bugs fixed in stretch:

- the completion code is completely broken (#855283)
- a member of the DSA team asked me to provide a systemd unit to avoid
  timeouts they encounter on porter boxes (#835104)

I'm very late with both of those issues that have been known for a long
time, I'm sorry for that.

unblock schroot/1.6.10-4

Here's the debdiff:

diff -Nru schroot-1.6.10/debian/changelog schroot-1.6.10/debian/changelog
--- schroot-1.6.10/debian/changelog 2017-01-09 18:38:50.0 +0100
+++ schroot-1.6.10/debian/changelog 2017-06-06 16:38:31.0 +0200
@@ -1,8 +1,17 @@
+schroot (1.6.10-4) unstable; urgency=medium
+
+  * Fix up bash completion file. Closes: #855283
+  * Add systemd service file to avoid issues with timeouts when
+you have many schroot sessions open. Closes: #835104
+Thanks to Laurent Bigonville for the patch.
+
+ -- Raphaël Hertzog   Tue, 06 Jun 2017 16:38:31 +0200
+
 schroot (1.6.10-3) unstable; urgency=medium
 
   * By default mark all mounts done by schroot-mount as "private"
 to avoid bad interactions caused by systemd's default of "shared"
-that resulted in failure to unmount them. Closes: #761435
+that resulted in failure to unmount them. Closes: #786566
 Thanks to Tyler Hicks  for the patch.
   * debian/patches/fix-test-suite-with-usrmerge.patch: fix test suite when
 /bin is a symlink to /usr/bin.
diff -Nru schroot-1.6.10/debian/control schroot-1.6.10/debian/control
--- schroot-1.6.10/debian/control   2017-01-09 18:38:50.0 +0100
+++ schroot-1.6.10/debian/control   2017-06-06 16:38:31.0 +0200
@@ -6,6 +6,7 @@
 Build-Depends:
  cmake (>= 2.8.12),
  debhelper (>= 9),
+ dh-systemd,
  pkg-config,
  libpam0g-dev,
  uuid-dev [!kfreebsd-any],
@@ -24,6 +25,7 @@
 Standards-Version: 3.9.8
 Vcs-Browser: https://anonscm.debian.org/cgit/buildd-tools/schroot.git
 Vcs-Git: https://anonscm.debian.org/git/buildd-tools/schroot.git
+Homepage: https://gitlab.com/codelibre/schroot
 
 Package: schroot-common
 Architecture: all
diff -Nru schroot-1.6.10/debian/patches/fix-bash-completion.patch 
schroot-1.6.10/debian/patches/fix-bash-completion.patch
--- schroot-1.6.10/debian/patches/fix-bash-completion.patch 1970-01-01 
01:00:00.0 +0100
+++ schroot-1.6.10/debian/patches/fix-bash-completion.patch 2017-06-06 
16:38:31.0 +0200
@@ -0,0 +1,25 @@
+Description: Fix bash completion
+Author: Raphaël Hertzog 
+Bug: https://gitlab.com/codelibre/schroot/issues/24
+Bug-Debian: https://bugs.debian.org/855283
+Last-Update: 2017-06-06
+---
+This patch header follows DEP-3: http://dep.debian.net/deps/dep3/
+--- a/etc/bash_completion/schroot
 b/etc/bash_completion/schroot
+@@ -15,7 +15,7 @@
+ # along with this program.  If not, see
+ # .
+ 
+-have schroot &&
++_have schroot &&
+ _schroot()
+ {
+ local cur prev options
+@@ -33,5 +33,4 @@ _schroot()
+ COMPREPLY=( $(compgen -W "$options" -- $cur) )
+ fi
+ return 0
+-}
+-complete -F _schroot schroot
++} && complete -F _schroot schroot
diff -Nru schroot-1.6.10/debian/patches/series 
schroot-1.6.10/debian/patches/series
--- schroot-1.6.10/debian/patches/series2017-01-09 18:38:50.0 
+0100
+++ schroot-1.6.10/debian/patches/series2017-06-06 16:38:31.0 
+0200
@@ -8,3 +8,4 @@
 fix-test-suite-with-usrmerge.patch
 Unmount-everything-that-we-can-instead-of-giving-up.patch
 fix-killprocs.patch
+fix-bash-completion.patch
diff -Nru schroot-1.6.10/debian/rules schroot-1.6.10/debian/rules
--- schroot-1.6.10/debian/rules 2017-01-09 18:38:50.0 +0100
+++ schroot-1.6.10/debian/rules 2017-06-06 16:38:31.0 +0200
@@ -25,7 +25,7 @@
 export DEB_CXXFLAGS_MAINT_APPEND := -Wall -D_FILE_OFFSET_BITS=64 
-D_LARGEFILE_SOURCE
 export DEB_BUILD_MAINT_OPTIONS := hardening=+all
 
-DH_OPTIONS = --buildsystem=cmake --builddirectory=debian/build --parallel
+DH_OPTIONS = --with systemd --buildsystem=cmake --builddirectory=debian/build 
--parallel
 
 ifneq (,$(filter parallel=%,$(DEB_BUILD_OPTIONS)))
   NUMJOBS = $(patsubst parallel=%,%,$(filter parallel=%,$(DEB_BUILD_OPTIONS)))
@@ -109,4 +109,7 @@
 override_dh_installinit:
dh_installinit --no-start --update-rcd-params='defaults'
 
+override_dh_systemd_start:
+   dh_systemd_start --no-start
+
 .PHONY: override_dh_auto_configure override_dh_auto_clean 
override_dh_auto_build override_dh_auto_test override_dh_auto_install 
override_dh_installchangelogs override_dh_installinit install-arch install-indep
diff -Nru schroot-1.6.10/debian/schroot.service 
schroot-1.6.10/debian/schroot.service
--- schroot-1.6.10/debian/schroot.service   1970-01-01 01:00:00.0 
+0100
+++ schro