[Cisco C4510R] CPU load van 40% gemiddeld normaal?

Pagina: 1
Acties:

Onderwerpen


Acties:
  • 0 Henk 'm!

  • alm
  • Registratie: September 2001
  • Laatst online: 20:51
Hoi,

Weet iemand of het normaal is dat een Cisco C4510R met Supervisor V blades (2 stuks in redundant mode) een constante gemiddelde CPU load van 40% heeft? Blade 3 heeft 1Gbps non-PoE en blade 4 t/m 9 hebben 1Gbps PoE poorten. Allemaal 48 poorts blades. Slot 10 is nog vrij. IOS versie is cat4000-i9s-mz.122-25.EWA14.

Er werd ons gezegd dat het niet goed is en dat we minder opties moeten gebruiken in de configuratie. Daar stonden aardig wat snmp zaken in en die heb ik behoorlijk verminderd, maar de load blijft zo rond de 40% hangen (dus geen verbetering of verslechtering). Ik heb nu geen idee wat ik er verder aan kan doen, behalve dan vragen of het normaal is of niet. Een zoektocht dmv Google en op de site van Cisco heeft helaas nog niks opgeleverd (wat normale cpu load is voor deze switch).

Overigens heeft een reload en het switchen naar een andere Supervisor ook geen effect gehad.

We hebben ook C3560's in gebruik en die zitten zo rond de 20% en C3750's die rond de 8% zitten. Beiden ipbase versies.

Acties:
  • 0 Henk 'm!

  • CyBeR
  • Registratie: September 2001
  • Niet online

CyBeR

💩

Ik heb een C3500XL die niets speciaals doet en altijd op 50% zit…

All my posts are provided as-is. They come with NO WARRANTY at all.


Acties:
  • 0 Henk 'm!

  • Jejking
  • Registratie: Februari 2005
  • Niet online

Jejking

It's teatime, doctor!

Dit gaat om servers, right? Ik ga er dan van uit dat het niet zo simpel is als Taakbeheer openen.

Acties:
  • 0 Henk 'm!

  • CyBeR
  • Registratie: September 2001
  • Niet online

CyBeR

💩

Nee, dit gaat niet om servers.

All my posts are provided as-is. They come with NO WARRANTY at all.


Acties:
  • 0 Henk 'm!

  • lordgandalf
  • Registratie: Februari 2002
  • Laatst online: 03-06 13:24
waar in je netwerk staat de 4510 als het een borderrouter is met BGp enz dan lijkt me 40% niet heel bijzonder imho. als het een simpel routertje of switchje is ergens (wat ik niet aanneem dan is het niet echt normaal maar goed zou het nog steeds kunnen.

Steam: Profile / Socialclub: Profile / Uplay: minedwarf / Origin: lordgandalf3


Acties:
  • 0 Henk 'm!

  • alm
  • Registratie: September 2001
  • Laatst online: 20:51
BGP staat niet aan, is onze coreswitch. Er hangen op andere plaaten in het gebouw nog wat andere switches aan vast (C3560's).

Acties:
  • 0 Henk 'm!

  • TrailBlazer
  • Registratie: Oktober 2000
  • Laatst online: 09:31

TrailBlazer

Karnemelk FTW

doe eens
sh processes cpu | e 0.0
dan zie je tenminste wat zoveel opeet

Acties:
  • 0 Henk 'm!

  • Predator
  • Registratie: Januari 2001
  • Laatst online: 20:12

Predator

Suffers from split brain

Ik heb even gekeken naar 2 cat 4506 hier.

De 2 staan in active/standby voor L3.
Er zit zéér weinig verschil tussen in CPU usage tussen de active and de standby, hoewel de standby veel minder (geen) L3 traffic krijgt.

De active zit op 35% CPU en de standby op 34% CPU. De CPU load verandert bijna niet.
Zelfs s'nachts als er weinig traffic is blijft die hetzelfde.

Dat is gewoon de MGMT van de CPU.

Als je ongeveer het volgende ziet is dat normaal:
code:
1
2
  40  14524496802808643699        517 15.27% 14.84% 14.88%   0 Cat4k Mgmt HiPri
  41  11533130601052626853       1095 14.79% 14.45% 13.41%   0 Cat4k Mgmt LoPri



Met 40% zou ik me dus echt nog geen zorgen gaan maken.

Ik heb net even op een 3560 gekeken en die draait ook 'idle' op 14% en die heeft absoluut niets te doen en heeft veel minder interfaces te managen dan een 4500...

Don't worry ... be happy ... :+
Jejking schreef op donderdag 06 mei 2010 @ 02:56:
Dit gaat om servers, right? Ik ga er dan van uit dat het niet zo simpel is als Taakbeheer openen.
D'oh 8)7 _O-

[ Voor 12% gewijzigd door Predator op 06-05-2010 09:50 ]

Everybody lies | BFD rocks ! | PC-specs


Acties:
  • 0 Henk 'm!

  • Ximon
  • Registratie: Juli 2004
  • Laatst online: 30-04 19:42
Misschien helpt undebug all. Als een vriendelijke collega z'n debug commando's niet op heeft geruimd na een projectje kan dat onnodig CPU cycles eten.

(╯°□°)╯︵ + ︵ x ︵ + ︵ x ︵ + ︵ x ︵ + ︵ x


Acties:
  • 0 Henk 'm!

Anoniem: 316920

Kun je iets meer vertellen over de taak van het beestje en hoe groot de distribution layer is (aantal routers/functies) ?

Ik heb wel eens core router setups gezien waarin veel security policy zaken werden geregeld. Deze hadden ook hoge loads en (merkbare) lage transfer rates. Ik neem aan dat dit hier niet het geval is ?

Acties:
  • 0 Henk 'm!

  • alm
  • Registratie: September 2001
  • Laatst online: 20:51
@TrailBlazer/Predator

code:
1
2
3
4
5
6
7
#sh processes cpu | exclude 0.0
CPU utilization for five seconds: 35%/1%; one minute: 38%; five minutes: 37%
 PID Runtime(ms)   Invoked      uSecs   5Sec   1Min   5Min TTY Process
  29    22968172  46421727        494 11.50% 11.41% 11.88%   0 Cat4k Mgmt HiPri
  30    37685420  31954416       1179 19.96% 22.69% 21.07%   0 Cat4k Mgmt LoPri
  57      239724   1242815        192  0.15%  0.12%  0.12%   0 IP Input
  94      135604  23772515          5  0.31%  0.32%  0.31%   0 PM Callback


Vergelijkbaar met die C4506 dus, alleen iets hoger totaal.

@Ximon:

Geen debugging wat loopt.

@Walt5011998:

Nee, doen weinig security zaken op de switch.

Acties:
  • 0 Henk 'm!

  • Predator
  • Registratie: Januari 2001
  • Laatst online: 20:12

Predator

Suffers from split brain

alm schreef op donderdag 06 mei 2010 @ 17:39:
@TrailBlazer/Predator

code:
1
2
3
4
5
6
7
#sh processes cpu | exclude 0.0
CPU utilization for five seconds: 35%/1%; one minute: 38%; five minutes: 37%
 PID Runtime(ms)   Invoked      uSecs   5Sec   1Min   5Min TTY Process
  29    22968172  46421727        494 11.50% 11.41% 11.88%   0 Cat4k Mgmt HiPri
  30    37685420  31954416       1179 19.96% 22.69% 21.07%   0 Cat4k Mgmt LoPri
  57      239724   1242815        192  0.15%  0.12%  0.12%   0 IP Input
  94      135604  23772515          5  0.31%  0.32%  0.31%   0 PM Callback


Vergelijkbaar met die C4506 dus, alleen iets hoger totaal.
Ik zou me er echt geen zorgen over maken, lijkt me normaal.

Everybody lies | BFD rocks ! | PC-specs


Acties:
  • 0 Henk 'm!

  • TrailBlazer
  • Registratie: Oktober 2000
  • Laatst online: 09:31

TrailBlazer

Karnemelk FTW

Ik vind het wel hoog als je het vergelijkt met bijvoorbeeld onze 6500 distributie switches. Ik heb alleen nul ervaring met die 4500 dus zal best goed zijn dan.

Acties:
  • 0 Henk 'm!

  • jvanhambelgium
  • Registratie: April 2007
  • Laatst online: 31-05 17:27
Bekijk dit eens

http://www.cisco.com/en/U...ote09186a00804cef15.shtml


Blijkbaar wat eigenlijk aan de architectuur van het beestje, tenzij je tegen die bug aanloopt, maar je IOS is wel recenter als die 12.1.(14)EW waar ze over spreken.


------------

Those process (Cat4k Mgmt HiPri, Cat4k Mgmt LoPri) are normally high
because of the way the CPU is managed on 4000/4500 technologies, let me
share with you the following information about it:
http://www.cisco.com/warp..._high_cpu.html#understand

This show processes cpu output shows that there are two processes that
use the CPU- Cat4k Mgmt HiPri and Cat4k Mgmt LoPri . These two processes
aggregate multiple platform-specific processes which perform the
essential management functions on the Catalyst 4500. These processes
process control plane as well as data packets that need to be
software-switched or processed.

In order to see which of the platform-specific processes use the CPU
under the context of Cat4k Mgmt HiPri and Cat4k Mgmt LoPri , issue the
show platform health command.

Each of the platform-specific processes has a target/expected
utilization of the CPU. When that process is within the target, the CPU
executes the process in the high-priority context. The show processes
cpu command output counts that utilization under Cat4k Mgmt HiPri . If
a process exceeds the target/expected utilization, that process runs
under the low-priority context. The show processes cpu command output
counts that additional utilization under Cat4k Mgmt LoPri . This Cat4k
Mgmt LoPri is also used to run background and other low-priority
processes, such as consistency check and reading interface counters.
This mechanism allows the CPU to run high-priority processes when
necessary, and the idle CPU cycles that remain are used for the
low-priority processes. To exceed the target CPU utilization by a small
amount, or a momentary spike in utilization, is not an indication of a
problem that needs investigation. However in our case its high. It seems
that you may have hit a software bug CSCdz50987. This bug was fixed with
release 12.1(14)EW.
-------------------------

Acties:
  • 0 Henk 'm!

  • DGTL_Magician
  • Registratie: Februari 2001
  • Laatst online: 14-05 19:00

DGTL_Magician

Kijkt regelmatig vooruit

alm schreef op donderdag 06 mei 2010 @ 17:39:
@TrailBlazer/Predator

code:
1
2
3
4
5
6
7
#sh processes cpu | exclude 0.0
CPU utilization for five seconds: 35%/1%; one minute: 38%; five minutes: 37%
 PID Runtime(ms)   Invoked      uSecs   5Sec   1Min   5Min TTY Process
  29    22968172  46421727        494 11.50% 11.41% 11.88%   0 Cat4k Mgmt HiPri
  30    37685420  31954416       1179 19.96% 22.69% 21.07%   0 Cat4k Mgmt LoPri
  57      239724   1242815        192  0.15%  0.12%  0.12%   0 IP Input
  94      135604  23772515          5  0.31%  0.32%  0.31%   0 PM Callback


Vergelijkbaar met die C4506 dus, alleen iets hoger totaal.

@Ximon:

Geen debugging wat loopt.

@Walt5011998:

Nee, doen weinig security zaken op de switch.
Ziet er prima uit hoor. Ik heb een aantal klanten 4500-series en die zitten ook in die richting.

Blog | aaZoo - (Wireless) Networking, Security, DDoS Mitigatie, Virtualisatie en Storage


Acties:
  • 0 Henk 'm!

  • Predator
  • Registratie: Januari 2001
  • Laatst online: 20:12

Predator

Suffers from split brain

TrailBlazer schreef op vrijdag 07 mei 2010 @ 07:25:
Ik vind het wel hoog als je het vergelijkt met bijvoorbeeld onze 6500 distributie switches. Ik heb alleen nul ervaring met die 4500 dus zal best goed zijn dan.
Geen appelen met peren vergelijken ;)

Een supervisor V voor de 4500 heeft maar een 400MHz cpu.
Ik denk dat er redelijk wat meer marge op een 6500 supervisor zit.
Dat mag ook wel gezien het verschil. :)

Everybody lies | BFD rocks ! | PC-specs


Acties:
  • 0 Henk 'm!

  • TrailBlazer
  • Registratie: Oktober 2000
  • Laatst online: 09:31

TrailBlazer

Karnemelk FTW

SUP720 heeft maar een 600 MHz CPU hoor
SR71000 CPU at 600Mhz, Implementation 0x504, Rev 1.2, 512KB L2 Cache

Wat ik tot een incident een tijdje geleden ook niet wist. De show proc cpu op een 6500 geeft de route processor output. Op een access switch doet die helemaal niks natuurlijk want L3 doet die niet zoveel. Veel interessanter is op zo'n device
remote command switch show proc cpu
Die stond namelijk wel op 100%

Acties:
  • 0 Henk 'm!

  • Predator
  • Registratie: Januari 2001
  • Laatst online: 20:12

Predator

Suffers from split brain

Dat is nog altijd 50% meer :P
Ik denk dat het verschil groter was.

Een Supervisor 6-E voor een 4500-E chassis draait op 1,3 Ghz.

Maar mhz is mhz niet natuurlijk.
De vergelijking gaat enkel op indien het dezelfde processor architechtuur is.

Everybody lies | BFD rocks ! | PC-specs


Acties:
  • 0 Henk 'm!

  • silvans
  • Registratie: Juni 2004
  • Laatst online: 03-06 23:03
Onze Supervisor 6 draait continu op 13%, dit neemt niet af of toe bij belasting van de switch. Prima dus :)

Acties:
  • 0 Henk 'm!

  • raymonvdm
  • Registratie: December 2001
  • Laatst online: 01-04 11:24
CPU utilization for five seconds: 17%/0%; one minute: 17%; five minutes: 19% ding gedraagd zich als een switch..

Acties:
  • 0 Henk 'm!

  • alm
  • Registratie: September 2001
  • Laatst online: 20:51
Die heb ik gelijk even opgeslagen in mijn favorieten! Bedankt voor de info! :)

@ALL:

Bedankt voor de input, ik ben gerust gesteld (niet dat ik er wakker van lag hoor, maar goed). Hoeven we tenminste niks aan de switch settings te doen. :)

[ Voor 21% gewijzigd door alm op 07-05-2010 20:50 ]

Pagina: 1