Talk:Magic SysRq key

Latest comment: 10 months ago by 2001:9E8:460D:E300:64FD:A42B:AA79:261A in topic REISUB

RSEIUB

edit

I readded the paragraph on RSEIUB, as there is a redirection from that acronym to this article and people searching for RSEIUB might not understand the relation to SysRq without it. (By the way: Why is it "and an incorrect one as well"?) --84.143.39.40 21:30, 12 January 2007 (UTC)Reply

RSEIUB is incorrect because S is the "Sync data to disk" command, and the E or I commands that signal applications to exit may cause the application to write more data to the disk, which would then be un-sync'd.

REISUB

edit

It's considered safer to sync after killing the processes so it might make sense to rewrite the mnemonic to Raising Elephants Is So Utterly Boring.

Perhaps one might even keep the "Skinny" part in place.

No changes made yet. —The preceding unsigned comment was added by 194.94.224.254 (talk) 08:33, 23 April 2007 (UTC).Reply

'It's considered safer'? Who considers it safer? Why would when you sync the disk matter? I think we need someone authoritative on this to say which one is better (if either) and why.--Cogburnd02 22:24, 17 August 2007 (UTC)Reply

I'm no authoritative reference, but if you sync before killing all processes, wouldn't they have a chance to queue more data to be written, which may be only partially written by the time the Unmount command is received? --Khumba 19:19, 29 August 2007 (UTC)Reply

Hmm... someone should ask a kernel developer which is best 75.70.81.208 11:15, 9 September 2007 (UTC)Reply

Processes can (and do) handle the SIGTERM signal. It's reasonable that a process that is asked to terminate gracefully will likely write log data (which might very well get cached). On the other hand, SIGKILL can't be handled, so doing a sync and then sending SIGKILL to all processes is safe. Andareed (talk) 06:44, 11 January 2008 (UTC)Reply

What if a process writes a file after the sync but before the SIGKILL? 2001:9E8:460D:E300:64FD:A42B:AA79:261A (talk) 19:54, 8 January 2024 (UTC)Reply

I'm no kernel hacker, but wouldn't remounting filesystems read-only in the end flush all the data anyway? --Piotr Mitas (talk) 13:51, 30 December 2008 (UTC)Reply

Explanation of REISUB commands

edit

When the article explains the actions taken by each command: REISUB, I is esplained kills all processes (except init), but before that its function is l sends the SIGKILL signal to all processes, including init., so, which one is right? the one that kills init or the one that doesn't? —Preceding unsigned comment added by 200.126.237.106 (talk) 16:22, 9 September 2007 (UTC)Reply

REISUB is at fault here. I removed it before, as it contradicts authoritative sources (the sysrq help file in the Linux sources), but it appears it was added again. I don't believe it appears in any reliable source. --Constantine 06:36, 4 November 2007 (UTC)Reply

Raising Skinny Elephants Is Utterly Boring

edit

"Raising Skinny Elephants Is Utterly Boring" redirects here, but is not explained in the article. I'm curious about what it means. -- Beland 02:51, 15 November 2007 (UTC)Reply

See this revision. -- intgr [talk] 03:59, 15 November 2007 (UTC)Reply

Enabling SysRq

edit

I think this info should have it's own section, or at the very least remain in 'Command line access and configuration' but be placed at the beginning and have the section renamed 'Configuration and command line access'. It just makes more sense to explain how to enable the feature before explaining advanced features which require the feature to already be activated. If there is agreement I'm happy to make the changes. -- (Kinslayer11 (talk) 04:14, 7 May 2008 (UTC))Reply

In my opinion this article has already crossed the "Wikipedia is not a manual" line, and should be trimmed rather than expanded. I believe this content is more appropriate on some Linux-oriented wiki like HowtoForge or others. -- intgr [talk] 11:37, 7 May 2008 (UTC)Reply

Is This Right?

edit

I'm confused. In the table, it seems to give conflicting info:
"Send the SIGTERM signal to all processes except init (PID 1)" = Qwerty key "e"
"Send the SIGKILL signal to all processes except init" = Qwerty key "i"
Is one supposed to quit all except init and the other including init? Or, do they both exclude init and maybe there's some difference between SIGTERM and SIGKILL that matters instead? I'm new to Linux, so maybe the article is right? If so, sorry~ ZeniffMartineau (talk) 12:24, 1 August 2010 (UTC)Reply

See SIGTERM, SIGKILL. -- intgr [talk] 09:39, 2 August 2010 (UTC)Reply
Thank you! Now it makes sense. I guess I didn't notice those links in the article somehow. :P ZeniffMartineau (talk) 07:11, 10 August 2010 (UTC)Reply

Power distribution unit

edit

Should it really be in "See Also" list? —Preceding unsigned comment added by 86.57.158.78 (talk) 18:02, 15 March 2011 (UTC)Reply

No, it shouldn't (I think). Removed it. PiusImpavidus (talk) 11:41, 5 April 2012 (UTC)Reply

REISUB Deletion

edit

Several people think the section on REISUB needs to be deleted. However, I disagree. I understand that Wikipedia should not be giving how-to instructions, but seriously, REISUB is the most common use of the SysRq key and some mention to it should be included. I agree the section previously included was much to instructional and should be re tweaked, but deleting it entirely (especially when REISUB redirects here) is the wrong way to do this.62.155.227.60 (talk) 17:56, 1 April 2011 (UTC)Reply

The Security concerns section

edit

I think the security concern section as it currently stands has some problems. From reading the section and the associated sources, it doesn't become clear to me how important the debate on the security issues on the SysRq key is, and if we, by including a section on it, are overrepresenting it. The second half of the section is rather how-to'ish. It leaves me wondering: 1. Should we include a section on the security in the first place? 2. If yes, what do we want it to include? Martijn Hoekstra (talk) 18:57, 6 April 2011 (UTC)Reply

Update (and History)

edit

It might be worth checking the history of the SysReq key as used in Linux. The keys used and files used for access have changed in purpose over the years. (For instance, one refered-to document - or source - mentiones /proc/sysrq-sticky which doesn't seem to exist in my current kernel).  DavidDouthitt  (Talk) 16:41, 3 February 2012 (UTC)Reply

How to use CTRL-ALT-F2 after SYSRQ would be a good example

edit

I agree that format of the current article is good. I hope the Wikipedia doesn't become like old fashioned printed encylopedias.

It would be useful to have a plausible example of how a person can un-stick a frozen system and get to a console without rebooting because often it's helpful to look at logs (like .xsession-errors) that will be overwritten if the system reboots.

Tashiro (talk) 00:50, 1 August 2012 (UTC)Reply

FreeBSD screenshot

edit

This article is talking about a Linux feature, but the kernel panic image near the end is FreeBSD.

If FreeBSD also supports magic SysRq in the same way (I do not believe it does), the article should be adjusted to clarify that. If not, a linux kernel panic image would be more appropriate. — Preceding unsigned comment added by 184.65.99.106 (talk) 04:22, 9 February 2013 (UTC)Reply

When the combination doesn't work.

edit

Last year my desktop computer (Fedora Linux, kept fully updated) started freezing. I activated the Magic SysRq Key and made sure it worked. However, it only worked when I didn't need it; that is, if the system froze, it wouldn't respond even to this. Eventually, I found that my CPU fan was dying and the freezes were caused by overheating. It should probably be mentioned in the article that hardware issues can prevent the system from responding to this, but all I have right now is a personal anecdote. Is this something that needs citation, or is it worth putting in as is?JDZeff (talk) 20:08, 20 September 2013 (UTC)Reply

I don't think so, as it is obvious that software (in systems like PCs, which are not specifically designed to be failsafe) is not garanteed to work if the underlying hardware is not working as it should. --Matthiaspaul (talk) 01:43, 21 September 2013 (UTC)Reply

laptops

edit

Hi, I added new method for laptops, I tesed it and it works(original laptop method mentioned in article doesn't work) please don't remove it. thanks, Alipoor90 (talk) 23:04, 14 December 2014 (UTC)Reply

I think it's only a matter of time before most of this article is written, because it fails WP:NOTHOWTO. -- intgr [talk] 06:47, 15 December 2014 (UTC)Reply

deleting REISUB entry

edit

My replacement of the REISUB entry by a mention that it's obsolete and sometimes actively harmful got reverted by an IP. I'm reverting back, not fishing for refs elsewhere -- I can point to [1] which is a self-quote, but one reviewed by multiple kernel maintainers. -KiloByte (talk) 09:34, 27 July 2022 (UTC)Reply

WRT previous discussions: I don't deny that REISUB was useful in the past. But that was then, and now is now. -KiloByte (talk) 20:31, 29 July 2022 (UTC)Reply