Procurve2524 CPU utilization problem

[LQrincz Zoltán]

I unpack my HP Procurve 2524 switch and just turn on. I checked from console(I did'nt use any other ethernet ports and modules) the boot process and I see the switch boot up well, but when I check the 'Status and Counters' and the 'General System Information' the CPU Utilization show me 93%. I don't know why?! The switch's fimware version is: 04.08. After 20 minutes the CPU utilization go down to 10%. In this period I don't use the switch. Somebody can answer why is high the CPU utilization after boot for 20 minutes. Thanks.

[Ron Kinner]

Latest firmware on 2524 is F.05.17 so I suggest you upgrade and then see if you still have the problem. See:

http://www.hp.com/rnd/software/switches.htm

Ron

[LQrincz Zoltán]

After the os upgrade to F_05_17 the CPU utilization go down to 90%. Not resolve the problem.

[Ron Kinner]

When the switch boots up it will try to find a DHCP/BOOTP server to get its management IP from. Have you tried setting the management IP? That should not be a big CPU user but you never know.

Ron

[LQrincz Zoltán]

On the IP configuration menu the IP config [DHCP/Bootp] is Disbaled. I have no idea why is working the CPU. Maybe the displaying is wrong.

[Luis Carrillo_1]

I have seen this on many of the 2524s we have. After a few hours they mellow out and only go down to about 50%. I do not trust the value, no customers are complaining. Seeing 99% almost made me take it out the first time I saw it.
Do all of them stay at 90%? if so I would take off nodes one by one until you find the one doing it.
If that doesnâ t work call support.

Good lu

[garrigues]

hi,

I just noticed that the same problem occurs on my 2524's. When I launch a big file transfer after a boot, CPU utilization rise from 90% to 99% and my telnet session is very slow. After the transfer, CPU utilization 'fall' to 90-92% until sysuptime is about 20mn (+/-5mn), then CPU suddenly get its 'normal' activity : 10-40%.

Upgrading from F.05.17 to F.05.22 firmware did not fix the problem.
Here are 2 config files used for my tests.
As you can see, access to DHCP isn't enabled, LACP and CDP are deactivated and i use VLANs;
ROM version is F.02.01, and these switches are connected to my network during my tests :


; J4813A Configuration Editor; Created on release #F.05.17

hostname "hp10"
snmp-server contact "Y/S"
snmp-server location "SR0"
time timezone 60
time daylight-time-rule Western-Europe
no cdp run
interface 1
no lacp
exit

(...)

interface 25
disable
exit
interface 26
disable
exit
ip default-gateway xxx.xxx.xxx.1
timesync sntp
sntp broadcast
snmp-server community "xxxxxxxx"
snmp-server community "xxxxxxx" Unrestricted
snmp-server host xxx.xxx.xxx.157 "public" All
vlan 1
name "DEFAULT_VLAN"
no ip address
no untagged 1-26
exit
vlan 96
name "INFO"
untagged 2-6,8-26
tagged 1,7
exit
vlan 2
name "ADMIN"
ip address xxx.xxx.xxx.80 255.255.255.128
tagged 1,7
exit
stack join 001083xxxxxx
no aaa port-access authenticator active
password manager
password operator

---------------------------------
; J4813A Configuration Editor; Created on release #F.05.22

hostname "hp3"
snmp-server contact "Y/S"
snmp-server location "RG"
time timezone 60
time daylight-time-rule Middle-Europe-and-Portugal
no cdp run
interface 1
no lacp
exit
interface 2
speed-duplex auto-10
no lacp
exit
interface 3
speed-duplex auto-10
no lacp
exit

(...)

interface 7
no lacp
exit
interface 8
speed-duplex auto-10
no lacp
exit

(...)

interface 25
no lacp
exit
interface 26
no lacp
exit
ip default-gateway xxx.xxx.xxx.1
timesync sntp
sntp broadcast
snmp-server community "xxxxxxxx"
snmp-server community "xxxxxxx" Unrestricted
snmp-server host xxx.xxx.xx.157 "public" All
vlan 1
name "DEFAULT_VLAN"
no ip address
no untagged 1-26
exit
vlan 96
name "INFO"
untagged 2-6,8-26
tagged 1,7
exit
vlan 2
name "ADMIN"
ip address xxx.xxx.xxx.13 255.255.255.128
tagged 1,7
exit
stack join 001083xxxxxx
primary-vlan 2
no aaa port-access authenticator active
password manager
password operator
----------------------------------

My others 2524 are in use and i cannot use them for testing purposes.

Do you know if this behavior noticeable for all the 2524 switches, and is it possible to fix this problem ?

Thanks !

sylvain;

[SCOOTER]

Sylvian,

This is normal behavior for the switch. Just wait about 15 min. after reloading the switch.

Regards,

SCOOTER

[Ardon]

Hi,

The reason why the CPU load is high during the first 20 min or so on a 2524 is normal as the CPU uses all idle cycles to generate SSH Session Keys (yes even if you are not using SSH). This is a trade off. Either generate them when an SSH connection is needed (long connection time a the Session Key has to be generated still) or generate them when there are idle CPU cycles anyways. HP decided for the latter.

Regards, Ardon

[OLARU Dan]

@Ardon

This cleared the air. What if I don't need SSH but use telnet instead? Can I disable the SSH stuff?