[j-nsp] EX4200 Junos 12.3R3.4 Processor Utilization

[Serge Vautour]

We hit an SNMP memory leak bug on 12.3R3 which drove up the CPU & Mem. Might be something similar. 12.3R4 is out. Maybe give it a try?

Serge




________________________________
 From: David Miller <dmiller at tiggee.com>
To: juniper-nsp at puck.nether.net 
Sent: Friday, September 20, 2013 5:59:35 PM
Subject: [j-nsp] EX4200 Junos 12.3R3.4 Processor Utilization
 


Anyone running the JTAC recommended version of Junos (12.3R3.4) on any
EX4200s and collecting configs with RANCID?  If so, then I would be
curious to see what your proc utilization looks like.

What I see on my 12.3R3.4 switch is a large CPU spike whenever RANCID
collects data.

The command that seems to be the issue is:
show version detail | no-more

On my switch the following 3 log messages appear in /var/log/messages
every time 'show version detail' is run:

Jun 14 03:19:39   mgd[2139]: UI_OPEN_TIMEOUT: Timeout connecting to peer
'remote-operations'
Jun 14 03:19:39   mgd[2139]: UI_OPEN_TIMEOUT: Timeout connecting to peer
'dhcp'
Jun 14 03:19:40   mgd[2139]: UI_OPEN_TIMEOUT: Timeout connecting to peer
'autoinstallation'

Top shows:
CPU states: 39.2% user,  0.0% nice, 55.3% system,  0.1% interrupt,  5.5%
idle
during a run of this show command.

My SNMP monitoring over time shows spikes of:
jnxOperatingCPU.FPC0  max: 81.80
jnxOperatingCPU.RE0   max: 35.92

Anyone else seeing this as well?

I can reproduce this behavior exactly on a brand new switch, upgraded to
12.3R3.4, with vanilla default Juniper config on it (so I don't think
that it is my config).

Am I missing some simple config option?
set chassis routing-engine dont-blow-out-cpus-on-simple-show-cmds

- DMM


_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp