number of stream pseudo terminals

[Jojo Castro]

Hi all,

Our systems operations has just escalated regarding terminals already consumed 80% of the total setting of nstrpty kernel param.

With this situation, is increasing the nstrpty online? Wha could be the effect if all the values has been consumed? Is there a proper procedure i can kill pts or terminals that are old and it think not in use?

Please see below data for your checking too.

Thanks in advance!


# kcusage -m nstrpty
Tunable: nstrpty
Setting: 60
Time Usage %
=============================================
Sat 04/25/09 33 55.0
Sun 04/26/09 31 51.7
Mon 04/27/09 36 60.0
Tue 04/28/09 35 58.3
Wed 04/29/09 32 53.3
Thu 04/30/09 35 58.3
Fri 05/01/09 30 50.0
Sat 05/02/09 31 51.7
Sun 05/03/09 33 55.0
Mon 05/04/09 32 53.3
Tue 05/05/09 30 50.0
Wed 05/06/09 33 55.0
Thu 05/07/09 32 53.3
Fri 05/08/09 29 48.3
Sat 05/09/09 37 61.7
Sun 05/10/09 42 70.0
Mon 05/11/09 43 71.7
Tue 05/12/09 39 65.0
Wed 05/13/09 33 55.0
Thu 05/14/09 31 51.7
Fri 05/15/09 32 53.3
Sat 05/16/09 32 53.3
Sun 05/17/09 32 53.3
Mon 05/18/09 32 53.3
Tue 05/19/09 37 61.7
Wed 05/20/09 39 65.0
Thu 05/21/09 39 65.0
Fri 05/22/09 38 63.3
Sat 05/23/09 35 58.3
Sun 05/24/09 35 58.3
Mon 05/25/09 50 83.3



# who -a | grep pts
esmscp pts/0 May 18 01:07 0:15 13240 10.123.45.141
esmscp pts/1 May 9 00:02 old 13393 10.123.45.141
. pts/2 Apr 1 00:28 . 15247 id= 2 term=0 exit=0
esmscp pts/3 May 12 15:08 . 9141 10.123.45.141
esmscp pts/4 May 12 15:09 0:19 509 10.123.45.141
. pts/5 Feb 21 02:55 . 16237 id= 5 term=0 exit=0
esmscp pts/6 May 17 19:53 . 26597 10.123.45.67
esmscp pts/7 May 17 19:53 0:06 27648 10.123.45.67
esmscp pts/8 May 9 18:24 old 4026 10.123.45.141
esmscp pts/9 May 25 09:34 old 27249 10.123.45.141
esmscp pts/10 May 12 15:14 0:15 5588 10.123.45.141
. pts/11 May 22 06:45 old 15737 id= 11 term=0 exit=0
esmscp pts/12 May 26 00:31 1:46 18990 10.123.45.141
. pts/13 Apr 23 07:35 old 13225 id= 13 term=0 exit=0
esmscp pts/14 May 7 10:09 old 982 10.123.45.141
esmscp pts/15 May 24 09:37 5:18 29262 10.123.45.141
esmscp pts/16 May 12 15:12 0:18 22627 10.123.45.141
esmscp pts/17 May 9 12:42 old 27102 10.123.45.141
. pts/18 Feb 4 05:01 old 426 id= 18 term=0 exit=0
esmscp pts/19 May 26 08:45 1:44 19877 10.123.45.139
esmscp pts/20 May 18 08:57 0:34 9286 10.123.45.141
. pts/21 Apr 19 14:12 . 15341 id= 21 term=0 exit=0
. pts/22 Feb 22 21:09 . 9678 id= 22 term=0 exit=0
. pts/23 Feb 24 06:33 . 12333 id= 23 term=0 exit=0
esmscp pts/24 May 26 08:44 1:45 19302 10.123.45.36
. pts/25 Mar 6 09:17 . 15452 id= 25 term=0 exit=0
. pts/26 Apr 18 14:53 . 18158 id= 26 term=0 exit=0
esmscp pts/27 May 26 02:05 1:46 6182 10.123.45.141
aab11661 pts/28 May 25 17:45 16:43 775 10.132.122.41
esmscp pts/29 May 25 07:22 old 18118 10.123.45.89
esmscp pts/30 May 25 07:55 . 4329 10.123.45.139
afl10746 pts/31 May 25 14:30 17:55 11835 10.132.122.70
. pts/32 May 22 09:33 . 3981 id= 32 term=0 exit=0
. pts/33 May 21 06:12 old 11579 id= 33 term=0 exit=0
. pts/34 May 24 08:05 17:55 20301 id= 34 term=0 exit=0
esmscp pts/35 May 25 17:24 0:03 20819 10.123.45.139
rgo07646 pts/36 May 25 18:17 16:12 11115 10.132.122.26
jsc4221 pts/37 May 25 18:51 . 932 10.123.45.100
rtc_ftp pts/38 May 25 18:56 15:33 22705 10.123.45.100
rtc_ftp pts/39 May 25 18:56 15:32 2890 10.123.45.100
. pts/40 May 26 10:20 0:09 9899 id= 40 term=0 exit=0
esmscp pts/41 May 26 09:54 0:34 12991 10.123.45.141
. pts/42 May 26 08:44 1:45 23298 id= 42 term=0 exit=0
esmscp pts/43 May 26 05:46 3:59 15985 10.123.45.36
. pts/44 May 26 05:51 4:38 5957 id= 44 term=0 exit=0
. pts/45 May 26 05:55 4:34 20060 id= 45 term=0 exit=0
esmscp pts/46 May 26 05:47 3:59 4875 10.123.45.36
esmscp pts/47 May 26 05:46 3:59 25538 10.123.45.36
esmscp pts/48 May 26 05:46 3:59 28156 10.123.45.36
. pts/49 May 26 05:52 4:37 2424 id= 49 term=0 exit=0

[Dennis Handly]

>With this situation, is increasing the nstrpty online?

Does kctune say nstrpty is dynamic?
It seems it isn't on 11.31:
http://docs.hp.com/en/B2355-60130/nstrpty.5.html

>Is there a proper procedure I can kill pts or terminals that are old and it think not in use?

Only what you make up. Are those sessions still accumulating CPU and have other processes other than the shell?

[Jojo Castro]

> Does kctune say nstrpty is dynamic?
It seems it isn't on 11.31:
http://docs.hp.com/en/B2355-60130/nstrpty.5.html

OS is 11.23 and it isn't dynamic.

> Only what you make up. Are those sessions still accumulating CPU and have other processes other than the shell?

Yes there are other processes. My question is from the output above, how can i kill those who are already runaway sessions? how can i determine those? is those with value term=0 exit=0?

[Dennis Handly]

>how can I kill those who are already runaway sessions? how can I determine those? is those with value term=0 exit=0?

What does "runaway" mean to you?
Those with "term=0 exit=0" may be the free ones. Perhaps kcusage(1m) gives you the max usage, not the current usage.

[Jojo Castro]

Hi,

I just instructed our operations to exit unnecessary sessions.