09-08-2015 07:15 AM - edited 03-08-2019 01:41 AM
Dear All,
I have a Cisco 7206VXR (NPE-G2) configured with BGP sessions with two adjacent neighbors and one gigabitethernet configured as gateway for my network.
Recently I noted a RTD peaks only on all IPs associated to router interfaces instead on pass through traffic latency is constant, please let me explain below:
Ping on G0/3 Gateway interface: Min = 0ms, Max = 250ms, Avg = 2ms
Ping on G0/1 BGP AS1 interface: Min = 0ms, Max = 266ms, Avg = 2ms
Ping on G0/2 BGP AS2 interface: Min = 0ms, Max = 248ms, Avg = 2ms
Ping on AS1 remote IP: Min = 0ms, Max = 8ms, Avg = 0ms
Ping on AS2 remote IP: Min = 7ms, Max = 38ms, Avg = 11ms
Ping on google dns: Min =17ms, Max = 38ms, Avg = 20ms
Response value of ping is suddenly high:
64 bytes from X.X.X.X: icmp_seq=94 ttl=255 time=0.252 ms
64 bytes from X.X.X.X: icmp_seq=95 ttl=255 time=0.202 ms
64 bytes from X.X.X.X: icmp_seq=96 ttl=255 time=0.371 ms
64 bytes from X.X.X.X: icmp_seq=97 ttl=255 time=0.247 ms
64 bytes from X.X.X.X: icmp_seq=98 ttl=255 time=0.175 ms
64 bytes from X.X.X.X: icmp_seq=99 ttl=255 time=0.153 ms
64 bytes from X.X.X.X: icmp_seq=100 ttl=255 time=0.192 ms
64 bytes from X.X.X.X: icmp_seq=101 ttl=255 time=0.253 ms
64 bytes from X.X.X.X: icmp_seq=102 ttl=255 time=0.260 ms
64 bytes from X.X.X.X: icmp_seq=103 ttl=255 time=0.203 ms
64 bytes from X.X.X.X: icmp_seq=104 ttl=255 time=0.256 ms
64 bytes from X.X.X.X: icmp_seq=105 ttl=255 time=226 ms
64 bytes from X.X.X.X: icmp_seq=106 ttl=255 time=0.260 ms
64 bytes from X.X.X.X: icmp_seq=107 ttl=255 time=0.272 ms
64 bytes from X.X.X.X: icmp_seq=108 ttl=255 time=0.172 ms
64 bytes from X.X.X.X: icmp_seq=109 ttl=255 time=0.207 ms
Traffic on router is always the same (between 150/200 Mbits), memory and cpu usage seems ok:
Head Total(b) Used(b) Free(b) Lowest(b) Largest(b)
Processor 538C940 835136308 535164112 299972196 298923988 298265396
I/O 38000000 67108864 4559756 62549108 62495984 62463516
Transient 37000000 16777216 44540 16732676 16207976 16732676
Maximum CPU% per second (last 60 seconds) -> 30%
Maximum CPU% per minute (last 60 minutes) -> 50%
Maximum CPU% per hour (last 72 hours) -> 60%
The issue doesn't seem associated to renew of ARP table, any suggestion? What I can check?
Thanks in advanced
Regards
Marco
09-08-2015 10:21 AM
Hi,
I don't believe there are any issues and what you're seeing is perfectly normal and expected.
You need to remember that IOS is a multi-tasking operating system that has many processes to handle. The way those processes are assigned resource within the CPU is controlled by the scheduler based upon the priority of the process.
From Inside Cisco IOS Software Architecture:
"IOS employs a priority scheme to schedule processes on the CPU. At creation time, every process is assigned one of four priorities based on the process’ purpose. The priorities are static; that is, they’re assigned when a process is created and never changed. The IOS process priorities are:
The way the scheduler works is to check and service all processes in the critical priority queue first, then move on to high, medium (or normal) and low priority queues in order. In between each scheduling of processes at some lower level, the next higher level queue is checked again to see if there are now any processes that need servicing at that higher level.
The priorities of each process can be seen with the show processes <pid> command. If you send ICMP traffic to a router then that traffic is handled by the IP Input process, which as seen below, has a Normal priority.
c1kv-1#sh proc 150 Process ID 150 [IP Input], TTY 0 Memory usage [in bytes] Holding: 53864, Maximum: 53864, Allocated: 0, Freed: 0 Getbufs: 0, Retbufs: 0, Stack: 46192/48000 CPU usage PC: 2C0D49A, Invoked: 20647, Giveups: 5, uSec: 28 5Sec: 0.00%, 1Min: 0.00%, 5Min: 0.00%, Average: 0.00% Age: 200895531 msec, Runtime: 594 msec State: Waiting for Event, Priority: Normal
This means that any ICMP traffic you send to the router will not be given resource on the CPU until there are no critical or high priority processes to be serviced. As such you can't guarantee or expect consistent response times as you have no way of determining which other more important processes are being serviced at any point in time.
The job of a router is to forward traffic and that task, and the processes required to undertake it, will always have priority over tasks such as responding to ICMP packets.
The post High latency issue in LAN.... talks of a similar issue. I've sure if you search this forum further you'll find more items like this raised previously, and the view is typically the same. Don't expect ping responses from a router to be consistent or any variance in response times to be an indication of a problem.
Regards
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide