07-13-2012 01:20 PM - edited 03-07-2019 07:45 AM
With Nikolay Karpyshev
Welcome to the Cisco Support Community Ask the Expert conversation. This is an opportunity to learn and ask questions about architecture and troubleshooting Access LAN Switches such as Cisco Catalyst 4500E, 3750-X, 3560-X, and 2960 with Cisco Expert Nikolay Karpyshev.
Nikolay Karpyshev is a Customer Support Engineer in the high touch technology support team (HTTS) at Cisco specialized in LAN Switching. Karpyshev supports the Cisco Switches Nexus 7000, Catalyst 6500, 3750, 3560, 4500, 2900, among others, and works as senior and escalation engineer. He was previously a part of Cisco Sales Associate program. He holds a specialist degree in Mathematics and Mechanics from Novosibirsk State University in Russia. Nikolay also holds these Cisco Certifications: CCNP, CCSP, and CCDP.
Remember to use the rating system to let Nikolay know if you have received an adequate response.
Nikolay might not be able to answer each question due to the volume expected during this event. Remember that you can continue the conversation on the Network Infrastructure sub-community discussion forum shortly after the event. This event lasts through July 27, 2012. Visit this forum often to view responses to your questions and the questions of other community members.
07-23-2012 10:45 PM
leolaohoo wrote:
Comments welcomed - even from you, Leo. :-)Thanks Darren.
3750X? In a DC? I'll agree with Nikolay. I would opt for 12.2(55)SE5. All my 3750X in the DC are running this IOS after I rolled back from 15.0(1)SE2. I don't think I'm brave enough to go 15.0(1)SE3 for 3750X in a DC.
Yes - 3750X in a DC. Unfortunately, I'm *very* budget constrained in this position, and am forced to use equipment which is not necessarily optimum for what we're doing. These things run routing (OSPF only, thankfully) and act as the switch infrastructure for a UCS and several other routers (2911's, which will be going away at the new DC, thankfully!), and they're the only way I can get wire-rate forwarding for the dark fibre links I need to run into the DC.
*I* asked them to buy me a couple of Nexus 3064's for the DC migration - and got soundly told to go away. :-)
Thanks to you and Nikolay - I'll grab 12.2.55 and upgrade to that before I migrate, to make sure everything is running the same IOS for the new stack.
Cheers.
07-23-2012 10:55 PM
Was running 15.0(1)SE2 and a week later the entire stack's log was filled with Traceback. I checked the config and got nothing "unique". The entire stack is a Layer 2 with 10 Gbps uplinks.
Anyway, rolled back to 12.2(55)SE5 and everything went away.
I didn't want to go to 15.0(1)SE1 because I know of the CPU hog issue. I am not too keen in 15.0(1)SE3 either because of past experience. I might consider 15.0(2)SE when this version comes out soon.
07-24-2012 04:50 AM
Hi Nikolay.
On a 4500 / SUP7; I tried to create a port-channel as a routed port with giga ethernet modules; but the "no switchport" command is rejected :
D7400_DC(config)#interface range Giga2/47 - 48
D7400_DC(config-if-range)#channel-group 2 mode active
Creating a port-channel interface Port-channel 2
D7400_DC(config-if-range)#exit
D7400_DC(config)#
D7400_DC(config)#Interface Port-channel 2
D7400_DC(config-if)# no shut
D7400_DC(config-if)# no switchport
Command rejected: Not a convertible port.
If i issue a "no switchport" on a single interface it works; but not on a bundle.
It seems that when I put the "no switchport" command on the physical interfaces and then on the port-channel it works.
Is it expected ?
regards.
07-24-2012 07:01 AM
Hi Surya,
Etherchanel inherits the features of L2 ports when it is configured with "channel-group" command on interface. When you do changes on port-channel interface it changes etherchannel config but not the config related to each individual physical port.
Thus trying to configure L3 on etherchanel via port-channel interface you creates the difference in etherchanel and physical port config. Thus it is blocked. When you configure it starting from physical port - it is then inherited by etherchanel and you can then apply that command to port-channel interface.
This is actually expected on all platforms at least Catalyst ones.
Nik
07-24-2012 10:15 AM
Can 3560-24PS be loaded with IOS 15 if it has sufficient flash memory (32MB)?
07-24-2012 03:00 PM
Can 3560-24PS be loaded with IOS 15 if it has sufficient flash memory (32MB)?
Yes you can but read Nikolay's response a few posts up.
07-24-2012 06:52 PM
Agree with Leo. There few things you also need to consider. Even if you have enough flash memory - the goal of IOS upgrade is to move from some defects affceting your network in current release or new features you need available only in the latest release.
If those are not the case - better to select one of the known stable release - e.g. 12.2(55)SE5.
Nik
07-25-2012 09:36 AM
Hello,
I am a student so patience please lol
Using packet tracer version 5.3.3, I set the switch to 2960 RSTP (spanning-tree mode rapid-PVST) and check with a PING from one host to another, I realized that the packet goes to the switch, it sends a broadcast and no longer a common unicast as stp.
This does not increase traffic on the network? large scale can become a big problem?
Sorry for the amateur! xD
07-25-2012 02:24 PM
Fernando,
Welcome to the community. Thanks for asking these questions. Nikolay will answer it. Just to let you know, we also have a community in Portuguese were you're welcome to participate. You can select Portuguese in the Language Pull down menu on top of the community page. You will be able to post questions also in your language. We're having a live webcast in Portuguese in August 21st.
Obrigada,
Monica Lluis
Global communities lead
07-25-2012 09:31 PM
Hi Fernando,
Let me check first if I understood your question correctly. You have two hosts connected to 2960 switch with RSTP on it. You do a ping from one host to another and capture packets with packet tracer. You see that switch sends a broadcast after receiving a ping from sender and then no unicast sent.
The broadcast possibly sent from host and not by switch. Before host can do a ping - it needs to resolve remote IP address to a MAC address if the destination IP is on the same subnet as the sender IP. To do it it uses ARP protocol and sends broadcast to all. Switch is receiving it checks and resends to all hosts within same VLAN.
So first broadcast is expected. If you don't see further unicast packets then it means that remote host for some reason does not send it's MAC address in response to ARP.
The process should be like this:
1. Diagram:
A--- switch ----B
To send ping to B A needs to know B MAC address. In the beginning A has on ip address so it needs to use ARP protocol to fond B MAC.
a. A sends ARP broadcast to switch to learn B MAC address
b. switch sends broadcast out of all ports within that VLAN except port to A
c. B receives that and sees it's own ip address and reply to A with unicast message toward the A MAC address it saw in broadcast packet source
d. Switch forwards unicast to A
- Thus if you see only single broadcast and nothing else - then be is either not receiving that broadcast or not responding to it.
If B - not receiving - check switch config to see if A and B are in same VLAN.
If B - not responding - check B for any config mismatch or other problems.
Let me know if I can further help you.
Nik
07-26-2012 05:38 AM
I understand. Thanks a lot.
Based on his explanation did some testing and realized what happens. You really helped me a lot.
I will return here several times to complete my journey to CCNA. xD
Again Thank you!
07-26-2012 06:52 AM
Glad it helped Fernando.
07-26-2012 05:54 AM
Hi Nikolay,
I am new to QoS and have a basic question on policing in 3560 switch.
Can you please explain how to calculate the burst-byte value for policing at ingress of 3560 switch? Is there any formula / cisco recommendations for calculating this value. I have studied that tokens are removed at 4000 times per second (interval) in 6500 switches, is there any such value for 3560 switches.
07-26-2012 06:58 AM
Hello,
The burst size define how much traffic you can send in single time interval (each second is device into several time slots depending on traffi demands). Burst you can send usually depends on the type of traffic and other factors. For the begining you can rely on Cisco recomended formula and then tune based on your needs:
normal burst = configured rate * (1 byte)/(8 bits) * 1.5 seconds
extended burst = 2 * normal burst
You can go through the links below to understand it in more depth
http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/polcing_shping_oview.html
http://www.cisco.com/en/US/docs/ios/12_2/qos/configuration/guide/qcfpolsh.html#wp1011592
Let me know if you have any further questions.
Nik
07-26-2012 08:16 AM
hi Nik,
if i have a Catalyst 3750x and i want to connect it to another brand of switch via Fiber optic link, will it work if i have a cisco SFP (1000Base-SX) on the Cat 3750x and a 1000Base-SX Fiber Transceiver on the 3Com switch? this really bothers me.
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