Hi, the output 'sh proc cpu | exc 0.00% 0.00% 0.00%' in both the 99% and "normal" usually helps a lot. :)
Calculate the difference for every process and find those that are significantly higher in the first case than in the second. If the differences don't add up to the difference of the totals (e.g. if you find that pretty much every process consumes the same amount of CPU in both the 99% and the normal case), then that means that the spike is caused by extensive interrupt activities, in other words packet switching. If this is the case, then you will have to go on a hunt to find the source of the packets that cause this spike. If the spikes come as regular as you seem to imply (once every hour?), then finding the pattern as to when they come might give a hint about their source. If you don't have any management tools that do this for you, then I would suggest to write a script that periodically polls CPU usage and stores the information with a time-stamp. Btw, what is your "normal" CPU usage? And how did you figure out that sometimes it's 99%? "Per-Second Jobs" is a generic timer process that wakes up every second and takes care of things that must be done every second. "IP Input" is the process handling the IP packets that are process switched. If you are running CEF, then this practically means the packets that are destined to the router itself (e.g. BGP). You might want to monitor the memory usage of these processes, and open a case with TAC if it is monotonically increasing (because that probably means that there is a memory leak). Although I am not quite sure how memory usage is related to CPU spikes. Thanks, Zsombor At 03:24 PM 6/27/2003 +0000, [EMAIL PROTECTED] wrote: >Yet another one of these post. > >I have a 7206vxr with an NPE 300 running 12.2.(13)T1. My CPU is spiking to >99% almost every hour yet I dont see any one process taking up large >amounts. My CPU is running as follows: > > >I ran the show tech-support through the Cisco interpreter and everything >expect some buffer errors (shown below): > > WARNING: This router has dropped 1070 packet(s) (0.09221%) due to a >shortage of > 'Small buffers'. > WARNING: This router has dropped 372 packet(s) (0.06801%) due to a >shortage of > 'Middle buffers'. > WARNING: This router has dropped 131 packet(s) (0.00586%) due to a >shortage of > 'Big buffers'. > WARNING: This router has dropped 103 packet(s) (1.03134%) due to a >shortage of > 'VeryBig buffers'. > ERROR: This router has dropped 102 packet(s) (80.95238%) due to a shortage >of > 'Large buffers'. > ERROR: This router has dropped 101 packet(s) (80.8%) due to a shortage of > 'Huge buffers'. > >....and memory utillization looked good. The output interpreter stated >that 2 processes are taking up more than 1 MB a piece which it is not >normal. I am running: CEF, BGP, OSPF, RIP, NBAR, ACLs and WCCP. Can >someone help me out in determining what these process are related to: > >Per-Second Jobs (Holding 1134364 bytes) >IP Input (Holding 1696336 bytes) > > >Perhaps direct me in how to go about tackling this issue since it is hard to >spot when it is happening since it does not stay peaked for very long. >Thanks. > > > > >Thanks, > >Mario Puras >SoluNet Technical Support >Mailto: [EMAIL PROTECTED] >Direct: (321) 309-1410 >888.449.5766 (USA) / 888.SOLUNET (Canada) Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=71549&t=71536 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
