1
00:00:07,050 --> 00:00:08,520
- All right, so we need to make sure

2
00:00:08,520 --> 00:00:11,280
that we have full sysrq functionality.

3
00:00:11,280 --> 00:00:16,280
That was in /proc/sys/kernel/sysrq.

4
00:00:18,270 --> 00:00:21,603
And we can see "1", and 1
is exactly what we need.

5
00:00:22,650 --> 00:00:26,100
Next, we need sysrq to
trigger a full OOM kill.

6
00:00:26,100 --> 00:00:28,200
I would always advise, before you do that,

7
00:00:28,200 --> 00:00:29,160
use echo h

8
00:00:29,160 --> 00:00:31,931
even if you're sure about
what you want to do.

9
00:00:31,931 --> 00:00:34,530
/proc/sysrq-trigger,

10
00:00:34,530 --> 00:00:36,030
and then dmesg

11
00:00:36,030 --> 00:00:38,250
to check what it is telling you.

12
00:00:38,250 --> 00:00:41,793
And there we can see
memory-full-oom-kill is f.

13
00:00:43,140 --> 00:00:44,760
The reason you need to check it out

14
00:00:44,760 --> 00:00:48,630
is that these sysrq-triggers
can do pretty nasty stuff

15
00:00:48,630 --> 00:00:49,980
on your system.

16
00:00:49,980 --> 00:00:52,020
Like, if by accident you type c,

17
00:00:52,020 --> 00:00:53,700
then you crash your system.

18
00:00:53,700 --> 00:00:55,140
And then it's gone.

19
00:00:55,140 --> 00:00:57,540
So you better be sure that you know

20
00:00:57,540 --> 00:00:59,970
what it is that you are going to do.

21
00:00:59,970 --> 00:01:04,970
So echo f > /proc/sysrq-trigger.

22
00:01:05,220 --> 00:01:08,190
And there we go, and again
it is killing the GDM,

23
00:01:08,190 --> 00:01:10,773
GNOME graphical Display Manager.

24
00:01:14,970 --> 00:01:17,610
Right. So now we are back.

25
00:01:17,610 --> 00:01:22,610
Let's have a look at
journalctl | grep -A 10 OOM,

26
00:01:25,950 --> 00:01:28,207
which is showing us, right here,

27
00:01:28,207 --> 00:01:32,160
"process of the unit has been
killed by the OOM killer."

28
00:01:32,160 --> 00:01:33,750
So we are running out of memory,

29
00:01:33,750 --> 00:01:35,010
and then the essence is

30
00:01:35,010 --> 00:01:37,470
that a process will be killed.

31
00:01:37,470 --> 00:01:39,540
And here you can see the information

32
00:01:39,540 --> 00:01:41,850
that was printed right after it.

33
00:01:41,850 --> 00:01:46,290
I have used -A on my grep, -A for "after."

34
00:01:46,290 --> 00:01:49,920
Maybe you wanna do minus
uppercase C for after and before.

35
00:01:49,920 --> 00:01:52,663
There's -A for after, -B for before,

36
00:01:52,663 --> 00:01:55,590
and -C for after and before,

37
00:01:55,590 --> 00:01:58,230
so that you can see exactly
what has been going on

38
00:01:58,230 --> 00:02:03,180
and get more information about
the process that was killed.

39
00:02:03,180 --> 00:02:05,760
In this case, by the way,
that was pretty obvious

40
00:02:05,760 --> 00:02:09,750
because we were kicked out of
the graphical Display Manager.

41
00:02:09,750 --> 00:02:13,380
In some other cases it
might not be that obvious.

42
00:02:13,380 --> 00:02:15,750
Hey, shall I show you something fun?

43
00:02:15,750 --> 00:02:20,613
Let's get back to this echo
h > /proc/sysrq-trigger,

44
00:02:21,840 --> 00:02:24,120
and dmesg.

45
00:02:24,120 --> 00:02:26,280
Because what else can we do?

46
00:02:26,280 --> 00:02:29,790
Well, we can do pretty dramatic stuff,

47
00:02:29,790 --> 00:02:34,790
like echo c > /proc/sysrq-trigger

48
00:02:38,580 --> 00:02:39,900
What is going to happen?

49
00:02:39,900 --> 00:02:42,390
Oh no, my system isn't
doing anything anymore.

50
00:02:42,390 --> 00:02:43,560
It crashes.

51
00:02:43,560 --> 00:02:45,720
That's what is going to happen.

52
00:02:45,720 --> 00:02:47,790
Now, you might wonder, why is this useful?

53
00:02:47,790 --> 00:02:49,770
Well, this is useful if you want to test

54
00:02:49,770 --> 00:02:54,000
what your system is
doing in specific cases.

55
00:02:54,000 --> 00:02:56,370
Maybe you have configured
some high availability,

56
00:02:56,370 --> 00:02:59,430
and you want stuff to be
started again automatically.

57
00:02:59,430 --> 00:03:02,070
Then isn't it cool that you
can test what is happening

58
00:03:02,070 --> 00:03:03,660
if your system is crashing.

59
00:03:03,660 --> 00:03:06,810
echo c > /proc/sysrq-trigger,

60
00:03:06,810 --> 00:03:08,970
and you got yourself a crash.

61
00:03:08,970 --> 00:03:11,550
And that is how cool, and
at the same time, dangerous

62
00:03:11,550 --> 00:03:13,323
the sysrq-triggers can be.