Your message dated Sun, 09 Oct 2022 19:51:32 +0000
with message-id <[email protected]>
and subject line Bug#1021474: fixed in runit 2.1.2-49
has caused the Debian Bug report #1021474,
regarding runit: forward signals from systemd-only services
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)


-- 
1021474: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1021474
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Package: runit
Version: 2.1.2-48
Severity: wishlist
X-Debbugs-Cc: [email protected]

Solutions listed in #1021465 provide access only to sysv service but there
are other issues left:

1. can't forward signals from systemd path, socket, timers; even an override
  of deb-systemd-invoke from i-s-h won't work as calls to deb-systemd-invoke
   in maintscript are run only when systemd is init.
2. same problem as above for packages that ship a systemd service without a
  sysvscript counterpart, either because it never existed or because it's
  been removed
3. the standard upgrade sequence that happens at postinstall, or at prerem
  and postrm is unfitted for runit design:
  - in runit, when a service is enabled is also automatically started, and
    when is disabled is also stopped, no way around it. So doing enable+ start
    at postinstall and stop + disable at pre|postrm is not just pointless, but
    can be racy like in #919296 
    Also, combo like 'enable but don't start' that are possible with debhelper
    are not feasible in runit.
  - the standard postinstall ( enable + rescan + start/restart ) sequence is not
    efficient when more services are involved in the same upgrade, for example 
if
    10 packages with services are upgraded the results is 
10*(enable+rescan+start/restart),
    while a more conveninent sequence could be 10*(enable) + 1*(rescan) + 
X*(restart),
    skipping the start of newly enabled service ( to avoid te race) and doing 
one
    rescan for the entire upgrade.
4. the one-to-one equivalence between runit and systemd or sysv services is not
  always feasible and sometimes is undesiderable:
  - runit does not have a service manager, supervised services are started in 
parallel
    and dependencies are automatically sorted out with polling; with this 
design splitting
    services into several pieces like systemd does is not a great idea as it 
makes relation
    between services more complex to solve
  - sometimes the opposite is true and one sysv or systemd service has to be 
expressed
    with more than one runit services ( for example the ExecStartPost systemd 
directive)

Overall, 1, 2 and 4 can be resolved by merging native runscript into packages 
but there
are other two problems:

* during the boot runit first runs all enabled sysv services in rc2.d (skipping 
when a native
  one exists), then starts all native services in parallel; so services that 
are involved in a
  dependency relationship should be merged in reverse order (otherwise the boot 
sequence will
  be broken). For example, before merging dbus runscript, all services that 
depends on
  dbus should be merged. This should be done without NMU as support for non 
default init
  is a whishlist bug. This is simply not feasible.
* there is a social issue with non default init in Debian; a certain group of 
maintainer will
  refuse to merge support for other init system.

So there is a need for a catch all 'runit-services' package at least for the 
above two case.
As a runit-services package ships only services definition while binaries are 
in separate packages
from other multiple sources, there is no established way to perform the 
standard maintscripts
sequence[1]. This is not just an issue of reduced support for an alternative 
init: silently
omitting to restart a runit service when a new version of the package that 
ships the binary
fixes an outstanding CVE is a bad, probably RC security bug.

An optimal solution is one that:

* provide enable/disable and restart when the service definition is in a 
separate
  package from the one the ships the binary (without cooperation from the 
latter)
* works when there a one-to-one relation with sysv or systemd services is not 
possible
  (basically decoupling runit services from other inits)
* provide an efficient sequence where services are first enabled/disabled, then 
runsvdir
  is told to rescan and finally only services that are upgraded are restarted
  (decoupled from dpkg configure order)
* bonus point if the mechanism can be used by local admin to handle their 
collection of
  services without the need to do a deb package

Probably it can be done by writing data that is usually in maintscirpt inside 
some file
that is parsed at runtime by a dpkg noawait trigger. This is kind of 
reinventing the wheel,
except that if it works, it's a technical solution to workaround a social 
problem.
Anyway I think this is within the scope of the "experimenting with alternative 
init" of the
last init GR.


[1] something like enable/disable the runit service when the package that ships 
the correspondent
    binary is installed/removed and restart the service on upgrade 



-- System Information:
Debian Release: bookworm/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (1, 'experimental')
merged-usr: no
Architecture: amd64 (x86_64)

Kernel: Linux 5.7.0-1-amd64 (SMP w/4 CPU threads)
Kernel taint flags: TAINT_FIRMWARE_WORKAROUND, TAINT_OOT_MODULE, 
TAINT_UNSIGNED_MODULE
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8), 
LANGUAGE=en_US:en
Shell: /bin/sh linked to /bin/dash
Init: runit (via /run/runit.stopit)
LSM: AppArmor: enabled

Versions of packages runit depends on:
ii  libc6           2.34-8
ii  runit-helper    2.14.2
ii  sysuser-helper  1.3.7+really1.4.1

Versions of packages runit recommends:
ii  runit-init  2.1.2-48

Versions of packages runit suggests:
ii  socklog  2.1.0+repack-4+b1
pn  zsh      <none>

-- Configuration Files:
/etc/default/runit changed [not included]
/etc/runit/ctrlaltdel changed [not included]
/etc/runit/runsvdir/single/sulogin/run [Errno 2] No such file or directory: 
'/etc/runit/runsvdir/single/sulogin/run'

-- no debconf information

-- debsums errors found:
debsums: changed file /lib/lsb/init-functions.d/40-runit (from runit package)
debsums: changed file /lib/runit/trigger_sv (from runit package)

--- End Message ---
--- Begin Message ---
Source: runit
Source-Version: 2.1.2-49
Done: Lorenzo Puliti <[email protected]>

We believe that the bug you reported is fixed in the latest version of
runit, which is due to be installed in the Debian FTP archive.

A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed.  If you
have further comments please address them to [email protected],
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
Lorenzo Puliti <[email protected]> (supplier of updated runit package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing [email protected])


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Format: 1.8
Date: Fri, 07 Oct 2022 23:27:21 +0200
Source: runit
Architecture: source
Version: 2.1.2-49
Distribution: experimental
Urgency: medium
Maintainer: Lorenzo Puliti <[email protected]>
Changed-By: Lorenzo Puliti <[email protected]>
Closes: 1021465 1021474
Changes:
 runit (2.1.2-49) experimental; urgency=medium
 .
   * make_svlink: create links also when CPSV_DIR is not /etc/sv
   * make the default directory for runsvdir configurable
   * stage 1: setup to experiment with a runtime service layout
   * /lib/lsb/init-functions.d/40-runit: mask sysv services and
      forward signals to runscripts; this can be tuned with flag files if
      start-stop-daemon wrapper is preferred. (Closes: #1021465)
   * add an sv_wtime utility needed for triggered upgrade of services
   * support triggered upgrade for runit services; this mode requires
      metafiles that are usually produced by dh-runit (Closes: #1021474)
   * invoke-run:
      - check for new metafile path inside the service directory
      - remove hardcoded path when 'chpst -e' is used
      - avoid loop when signals for sysv script are forwarded
   * cpsv:
      - improve readability of -d with colored diff
      - improve heuristic on sysvinit scripts
   * update-service:
      - add policy compliant commands
      - update manpage
   * run_sysv_scripts: small optimizations
   * svlogd: copy pkg metafile inside the log directory
   * Add a purge fallback for runit package, to fix piuparts
      sporadic failure
Checksums-Sha1:
 844f7daf8a4cfec2f0796ea5328d2d1a95667a57 2274 runit_2.1.2-49.dsc
 a07cb454c55b79648c3a9210b597a3d1eebb5be9 59372 runit_2.1.2-49.debian.tar.xz
 a499d76e2b2e29b9621797550de6cd186ffd1233 5898 runit_2.1.2-49_source.buildinfo
Checksums-Sha256:
 656a686151e9b8379d54aabaf00bb783d69636a4adccf989e153d9d74db3fcbe 2274 
runit_2.1.2-49.dsc
 df372f8958714abc2a093c13f918e3c675497daee31ed54a8087db9dce1a8946 59372 
runit_2.1.2-49.debian.tar.xz
 2465173e8d3deb93ba732d657bc39db9060b085ace5979d1162ef39e511fe72b 5898 
runit_2.1.2-49_source.buildinfo
Files:
 709005d56c7ab1610d72028a6f268b03 2274 admin optional runit_2.1.2-49.dsc
 87eacfa074b2b5c6f3e7a1a92181b946 59372 admin optional 
runit_2.1.2-49.debian.tar.xz
 344d2288f06a0c68f0b4168491f4ca20 5898 admin optional 
runit_2.1.2-49_source.buildinfo

-----BEGIN PGP SIGNATURE-----
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=euDA
-----END PGP SIGNATURE-----

--- End Message ---

Reply via email to