On 23.02.2018 09:59, Peter Suetterlin wrote:

Per Jessen wrote:

...

I thought I would do a full reinstall this time - just server pattern on a xen guest. I noticed the partitioner did not automatically re-use an existing swap parittion, but opted to carve a new one out.

I'd call it the safe way. Could be the existing one contains a hibernated system?

This can be detected easily: the signature is different then. -- Stefan Seyfried Ceterum censeo fluid-soundfont esse delendam (from the Leap 15 DVD :-) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On 2018-02-23 09:59, Peter Suetterlin wrote:

Per Jessen wrote:

...

I thought I would do a full reinstall this time - just server pattern on a xen guest. I noticed the partitioner did not automatically re-use an existing swap parittion, but opted to carve a new one out.

I'd call it the safe way. Could be the existing one contains a hibernated system?

Formatting the swap partition of an hibernated system will cause it to crash, and corrupt all the partitions it was using. Normally, a computer that is hibernated will not allow to boot any other system, including the install DVD/stick. Although it can be forced to do boot another system, this is very dangerous. If the partitioner finds an hibernation image it should abort. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)

On Fri, 23 Feb 2018 14:52:15 +0100 Peter Suetterlin wrote:

Nitpicking: It's not running, so it can hardly crash :D

Exactly. I have installed a number of multi-boot PCs with Windows and 3,4+ Linux distros, all the distros sharing a single swap partition and a single /home partition (with different user accounts). It works fine. If you try to boot a different Linux, it finds someone else's signature in the swap partition, discards the saved image and boots normally, as far as I can recall. Windows, of course, doesn't know or care. If you try to resume a hibernated OS and its hibernation image has gone, it just boots normally. The only problem I have encountered is a new one and recent. Windows 10 will sometimes hibernate when you ask it to shut down, so that it starts quicker next time. If I do not know this and boot Linux, systemd will not mount the Windows partitions because they were not cleanly unmounted -- and since my Windows partitions are automounted at boot time, and contain my work folders, this puts systemd in a loop, endlessly retrying to mount unclean C: and D: NTFS volumes. The only solution is to boot Windows and shut it down properly. The snag is, systemd *also* does this if it can't mount my Linux partitions for some reason, and then, the solution is to just force a reboot with Ctrl-Alt-Del. But it is not visible appearent _which_ partitions won't mount without careful inspection, so normally, I reboot a few times before I figure out Windows is to blame. This is the one thing about systemd that really annoys me.

While yes, it's likely leaving unclean partitions, I'd expect the hibernate script also to do something like a sync before, so damage should be small.

I think it does.

...

Normally, a computer that is hibernated will not allow to boot any other system, including the install DVD/stick.

What? TBH, this is the first time I ever hear this. And I *have* booted hibernated systems with a different OS. Quite some time ago though - I'm exclusively using suspend nowadays.

I agree. This is news to me too, and does not match what I have seen on my own machines. -- Liam Proven - Technical Writer, SUSE Linux s.r.o. Corso II, Křižíkova 148/34, 186-00 Praha 8 - Karlín, Czechia Email: lproven@suse.com - Office telephone: +420 284 241 084 -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

On Fri, 23 Feb 2018 16:13:35 +0100 Knurpht - Gertjan Lettink wrote:

This can be worked around by disabling FastBoot in Windows. That is what leaves the FS in an unclean state.

Thanks for the info. I will find out how to do that. -- Liam Proven - Technical Writer, SUSE Linux s.r.o. Corso II, Křižíkova 148/34, 186-00 Praha 8 - Karlín, Czechia Email: lproven@suse.com - Office telephone: +420 284 241 084 -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Liam Proven composed on 2018-02-23 16:06 (UTC+0100):

The only problem I have encountered is a new one and recent.

Windows 10 will sometimes hibernate when you ask it to shut down, so that it starts quicker next time. If I do not know this and boot Linux, systemd will not mount the Windows partitions because they were not cleanly unmounted -- and since my Windows partitions are automounted at boot time, and contain my work folders, this puts systemd in a loop, endlessly retrying to mount unclean C: and D: NTFS volumes.

The only solution is to boot Windows and shut it down properly.

Another besides disabling FastBoot in Windows is to avoid boot dependence on the Windows volumes via the nofail fstab option.

The snag is, systemd *also* does this if it can't mount my Linux partitions for some reason, and then, the solution is to just force a reboot with Ctrl-Alt-Del.

But it is not visible appearent _which_ partitions won't mount without careful inspection, so normally, I reboot a few times before I figure out Windows is to blame.

I'm pretty sure that when using nofail, dmesg will report which failed to mount.

This is the one thing about systemd that really annoys me.

Many more than one here. e.g. chkconfig -l output would always be alphabetized and neatly fit the scrollback buffer if not a single screen. -- "Wisdom is supreme; therefore get wisdom. Whatever else you get, get wisdom." Proverbs 4:7 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 El 2018-02-23 a las 14:52 +0100, Peter Suetterlin escribió:

Carlos E. R. wrote:

...

On 2018-02-23 09:59, Peter Suetterlin wrote:

...

Per Jessen wrote:

...

I thought I would do a full reinstall this time - just server pattern on a xen guest. I noticed the partitioner did not automatically re-use an existing swap parittion, but opted to carve a new one out.

I'd call it the safe way. Could be the existing one contains a hibernated system?

Formatting the swap partition of an hibernated system will cause it to crash, and corrupt all the partitions it was using.

Nitpicking: It's not running, so it can hardly crash :D

Well, it not be able to thaw at all.

While yes, it's likely leaving unclean partitions, I'd expect the hibernate script also to do something like a sync before, so damage should be small. Better than just switching off the power/pressing reset, which most systems handle quite well these days.

No, hibernate does not sync the filesystems. But I was mistaken: erasing the swap partition in that case is similar to hitting the power button.

...

Normally, a computer that is hibernated will not allow to boot any other system, including the install DVD/stick.

What? TBH, this is the first time I ever hear this. And I *have* booted hibernated systems with a different OS. Quite some time ago though - I'm exclusively using suspend nowadays.

You have been lucky. Situation. You have a running system, and it hibernates. The filesystems remain open, files half written, some structures in ram, which are swapped, not flushed. Then you boot a second system in the same machine, and it mounts some or all the "hibernated" partitions. It thinks that the partitions are dirty, not cleanly umounted (as if the power button was hit), so it fscks the filesystems. Perhaps some files are written to those. Then thys system powers off. Next the first filesystem thaws. It thinks that the filesystems are in the state *it* left them, but they aren't. It writes the in memory structures to disk *in its previous unrepaired status*. The resulting destruction on those filesystems is epic. It happened to me, just once, so I know.

...

If the partitioner finds an hibernation image it should abort.

NACK, by all means.

I might very well know what I'm doing. Just aborting and refusing to work is exactly the kind of Windows-typical behavior that I do **NOT** want to see on Linux.

A clear warning is fine. Everything else is not

Nope. Abort. :-) If you know what you are doing, issue a command line undocumented option that allows you to continue. - -- Cheers Carlos E. R. (from openSUSE 42.3 x86_64 "Malachite" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iF4EAREIAAYFAlqQU+QACgkQja8UbcUWM1y+9QEAjNN++7Zv3sXWNjkcpEEchbjS DFY1hJ6nR3iDXm5dIbIBAIPwHWhCogEW8Pal5XIAJgKGjIrfz3XzfHIrDcTmSshN =SzLq -----END PGP SIGNATURE-----