[opensuse] Leap 42.2 - Why is systemd invoked with undocumented --switched-root and --deserialize options?
All, Leap 42.2 uses both --switched-root and --deserialize which are *intentionally* undocumented features. See: http://lists.freedesktop.org/archives/systemd-devel/2013-September/013008.ht... Why are they used in Leap 42.2? On my Leap 42.2, I see systemd running with: /usr/lib/systemd/systemd --switched-root --system --deserialize 25 Is this intended? -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/02/2017 07:34 PM, David C. Rankin wrote:
--switched-root --system --deserialize 25
Google found this: http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l... Presumable you may have resumed from suspend or something? My Leap machine shows the same, but my Manjaro machine does not. It doesn't suspend because it has a VM running a service daemon. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/02/2017 08:01 PM, John Andersen wrote:
On 01/02/2017 07:34 PM, David C. Rankin wrote:
--switched-root --system --deserialize 25
Google found this:
http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l...
Presumable you may have resumed from suspend or something? My Leap machine shows the same, but my Manjaro machine does not. It doesn't suspend because it has a VM running a service daemon.
Nope. A boot from power off showed the same thing. Open source but undocumented. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/02/2017 10:07 PM, John Andersen wrote:
Nope. A boot from power off showed the same thing. Open source but undocumented.
Yep, and for the curious, that means picking through the open-source to find out why and the hell SuSE uses the open but undocumented features. I'm sure there is a sound reason -- I just want to know what the hell it is.... Inquiring minds want to know.... It sucks... -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 3 January 2017 at 07:14, David C. Rankin
On 01/02/2017 10:07 PM, John Andersen wrote:
Nope. A boot from power off showed the same thing. Open source but undocumented.
Yep, and for the curious, that means picking through the open-source to find out why and the hell SuSE uses the open but undocumented features. I'm sure there is a sound reason -- I just want to know what the hell it is....
Inquiring minds want to know.... It sucks...
Then Inquire in the correct place? opensuse-factory@opensuse.org would be more appropriate than here Or more sensibly you could look into OBS and find the email address of the person maintaining it - https://build.opensuse.org/package/view_file/openSUSE:Leap:42.2/systemd/syst... Asking them directly is likely to get you the most accurate answer as to why. Good luck ;) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Richard Brown wrote:
Then Inquire in the correct place? opensuse-factory@opensuse.org would be more appropriate than here
Or more sensibly you could look into OBS and find the email address of the person maintaining it - https://build.opensuse.org/package/view_file/openSUSE:Leap:42.2/systemd/syst...
Asking them directly is likely to get you the most accurate answer as to why. Good luck ;)
---- I don't think the suse people will necessarily know, as it is systemd that uses those arguments to "talk to itself". Or, more accurately, another copy of its self. It seems it is related to moving systemd's "root" from the ramdisk to the harddisk, thus the "switched-root" flag. The deserialize tells the new systemd where to read in its state information. I may be wrong, but the number following it may be a file descriptor. I find it a bit odd that they don't document those things, since by not doing so, they draw attention to "internal switches" that they don't want to "advertise". I guess they haven't gotten the news: security through obscurity...isn't. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 On 2017-01-03 19:39, L A Walsh wrote:
Richard Brown wrote:
I find it a bit odd that they don't document those things, since by not doing so, they draw attention to "internal switches" that they don't want to "advertise". I guess they haven't gotten the news: security through obscurity...isn't.
It reminds me of Microsoft. They have been doing that for ages, and then people making money by publishing "Undocumented Windows functions, Number Eleven". But systemd devs are not the first to do this. Other opensource projects had undocumented switches for many years. I remember one that named those switches "use-the-force-luke" or something similar. - -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) iF4EAREIAAYFAlhr9r0ACgkQja8UbcUWM1yP6QD/aZiT8xwwDwYksgF8xx+gG/WK zwQ39abf3iTCjH70owIBAIL2fcgteYBirqruYSQwZGg3cZ1o/klz1yiRcgI6jTeH =TdNM -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/03/2017 11:08 AM, Carlos E. R. wrote:
But systemd devs are not the first to do this. Other opensource projects had undocumented switches for many years.
Systemd may have such switches for various things, bug tracing harness, etc. But when you find it used in a production distro you just HAVE to wonder WTF? If Opensuse is going to use this it becomes THEIR problem to document it, even if systemd in their arrogance decide not to. None of my other machines are using this except one: Solus/Budgie. -- _____________________________________ ---This space for rent---
Hello! Seems, a brief description is here: "These are intentionally undocumented internal parts of systemd. Very simply, therefore: --deserialize is used to restore saved internal state that a previous invocation of systemd, exec()ing this one, has written out to a file. Its option argument is an open file descriptor for that process. --switched-root is used to tell this invocation of systemd that it has been invoked from systemd managing an initramfs, and so should behave accordingly — including turning off some of the behaviour otherwise caused by --deserialize." Source: http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l... 03.01.2017 23:25, John Andersen пишет:
On 01/03/2017 11:08 AM, Carlos E. R. wrote:
But systemd devs are not the first to do this. Other opensource projects had undocumented switches for many years. Systemd may have such switches for various things, bug tracing harness, etc. But when you find it used in a production distro you just HAVE to wonder WTF?
If Opensuse is going to use this it becomes THEIR problem to document it, even if systemd in their arrogance decide not to.
None of my other machines are using this except one: Solus/Budgie.
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/03/2017 2:11 PM, Mikhail Kasimov wrote:
Source: http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l...
Yes, already posted upstream. Still not at all helpful, and not a single word of explanation as to why this needs to remain undocumented. Nor is there any indication why opensuse does it this way where most distros don't. -- _____________________________________ ---This space for rent--- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 03/01/17 22:15, John Andersen wrote:
On 01/03/2017 2:11 PM, Mikhail Kasimov wrote:
Source: http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l...
Yes, already posted upstream.
Still not at all helpful, and not a single word of explanation as to why this needs to remain undocumented.
Nor is there any indication why opensuse does it this way where most distros don't.
Because not all distros default to using initramfs? My main systems run OpenRC, so I can't do any testing on that front, but they also run mirrors so they HAVE TO boot from initramfs. I'm guessing if they ran systemd they would behave this way. I doubt OpenSUSE are intentionally doing anything at all. My immediate reaction on seeing David's original post was "is this a side effect of an initramfs?" So as somebody who doesn't even use systemd, I guessed almost instantly what appears to be going on. It's nothing to do with SUSE - and everything to do with systemd. (I would add, I have a similar situation with my OpenRC systems, one of which doesn't work correctly, so I have investigated this grub/boot/linux area reasonably well ... :-) Cheers, Wol -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/04/2017 11:56 AM, Wols Lists wrote:
On 03/01/17 22:15, John Andersen wrote:
On 01/03/2017 2:11 PM, Mikhail Kasimov wrote:
Source: http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l...
Yes, already posted upstream.
Still not at all helpful, and not a single word of explanation as to why this needs to remain undocumented.
Nor is there any indication why opensuse does it this way where most distros don't.
Because not all distros default to using initramfs?
My main systems run OpenRC, so I can't do any testing on that front, but they also run mirrors so they HAVE TO boot from initramfs. I'm guessing if they ran systemd they would behave this way.
I doubt OpenSUSE are intentionally doing anything at all. My immediate reaction on seeing David's original post was "is this a side effect of an initramfs?" So as somebody who doesn't even use systemd, I guessed almost instantly what appears to be going on. It's nothing to do with SUSE - and everything to do with systemd.
(I would add, I have a similar situation with my OpenRC systems, one of which doesn't work correctly, so I have investigated this grub/boot/linux area reasonably well ... :-)
Cheers, Wol
So your best answer is that I mistook the fact that these other distros were using systemd when they actually weren’t? Thanks for your vote of confidence. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 04/01/17 20:08, John Andersen wrote:
On 01/04/2017 11:56 AM, Wols Lists wrote:
On 03/01/17 22:15, John Andersen wrote:
On 01/03/2017 2:11 PM, Mikhail Kasimov wrote:
Source: http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l...
Yes, already posted upstream.
Still not at all helpful, and not a single word of explanation as to why this needs to remain undocumented.
Nor is there any indication why opensuse does it this way where most distros don't.
Because not all distros default to using initramfs?
My main systems run OpenRC, so I can't do any testing on that front, but they also run mirrors so they HAVE TO boot from initramfs. I'm guessing if they ran systemd they would behave this way.
I doubt OpenSUSE are intentionally doing anything at all. My immediate reaction on seeing David's original post was "is this a side effect of an initramfs?" So as somebody who doesn't even use systemd, I guessed almost instantly what appears to be going on. It's nothing to do with SUSE - and everything to do with systemd.
(I would add, I have a similar situation with my OpenRC systems, one of which doesn't work correctly, so I have investigated this grub/boot/linux area reasonably well ... :-)
Cheers, Wol
So your best answer is that I mistook the fact that these other distros were using systemd when they actually weren’t?
If you actually read what I wrote, I never said anything of the sort. It's the COMBINATION of initramfs and systemd. I don't know whether these other distros use an initramfs - I certainly avoid it if I can. I just don't have any choice.
Thanks for your vote of confidence.
Cheers, Wol -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/04/2017 12:33 PM, Wols Lists wrote:
On 04/01/17 20:08, John Andersen wrote:
On 01/04/2017 11:56 AM, Wols Lists wrote:
On 03/01/17 22:15, John Andersen wrote:
On 01/03/2017 2:11 PM, Mikhail Kasimov wrote:
Source: http://unix.stackexchange.com/questions/98128/what-are-the-systemd-command-l...
Yes, already posted upstream.
Still not at all helpful, and not a single word of explanation as to why this needs to remain undocumented.
Nor is there any indication why opensuse does it this way where most distros don't.
Because not all distros default to using initramfs?
My main systems run OpenRC, so I can't do any testing on that front, but they also run mirrors so they HAVE TO boot from initramfs. I'm guessing if they ran systemd they would behave this way.
I doubt OpenSUSE are intentionally doing anything at all. My immediate reaction on seeing David's original post was "is this a side effect of an initramfs?" So as somebody who doesn't even use systemd, I guessed almost instantly what appears to be going on. It's nothing to do with SUSE - and everything to do with systemd.
(I would add, I have a similar situation with my OpenRC systems, one of which doesn't work correctly, so I have investigated this grub/boot/linux area reasonably well ... :-)
Cheers, Wol
So your best answer is that I mistook the fact that these other distros were using systemd when they actually weren’t?
If you actually read what I wrote, I never said anything of the sort. It's the COMBINATION of initramfs and systemd. I don't know whether these other distros use an initramfs - I certainly avoid it if I can. I just don't have any choice.
Thanks for your vote of confidence.
Cheers, Wol
Ive actually looked closely at some of the ones that do not use the opensuse way and they appear to be invoking systemd via /sbin/init which is linked to /lib/systemd/systemd. This is documented as an compatibility method of invocation of systemd. Some of these distros ALSO offer a sysvinit option so its probably just one less thing to keep maintenance easier. However it does change the structure of the running system considerably and systemd (under the disguise of /sbin/init) does not run those command line arguments. On these systems, (Such as Manjaro, Raspberian), the command sudo ps axjf |less shows a whole bunch of things (about 200 items) which are running subordinate to kthread (pid 2), and /sbin/init is pid 1. Both of these use initramfs. -- After all is said and done, more is said than done. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 01/03/2017 10:39 AM, L A Walsh wrote:
It seems it is related to moving systemd's "root" from the ramdisk to the harddisk, thus the "switched-root" flag. The deserialize tells the new systemd where to read in its state information. I may be wrong,
It sounds like that, but Arch boots from ramdisks as well, and I never see this running that way. -- _____________________________________ ---This space for rent--- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (7)
-
Carlos E. R.
-
David C. Rankin
-
John Andersen
-
L A Walsh
-
Mikhail Kasimov
-
Richard Brown
-
Wols Lists