This discussion is locked

ASK THE EXPERT - Nexus 5000 & 2000 series

Unanswered Question
Feb 25th, 2011

with Lucien Avramov

Welcome to the Cisco Networking Professionals Ask the Expert conversation. This is an opportunity to learn how to troubleshoot Nexus 5000 & 2000 series with Lucien Avramov.  Lucien is a customer support engineer at the Cisco Technical  Assistance Center. He currently works in the data center switching team supporting customers on the Cisco Nexus 5000 and 2000. He was previously a technical leader within the network management team. Lucien holds a  bachelor's degree in general engineering and a master's degree in  computer science from Ecole des Mines d'Ales. He also holds the  following certifications: CCIE #19945 in Routing and Switching, CCDP,  DCNIS, and VCP #66183.

Remember to use the rating system to let Lucien know if you have received an adequate response.

Lucien 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 Security discussion forums shortly after the event. This event lasts through March 11, 2011. Visit this forum often to view responses to your questions and the questions of other community members.

I have this problem too.
2 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Average Rating: 5 (5 ratings)
wu.kristi Wed, 03/02/2011 - 10:02

Lucien

I have a 2232 and I am not able to configure it. I am running NX-OS 4.1(3)N1(1). Can you help? Thanks.

Lucien Avramov Wed, 03/02/2011 - 11:57

I'm glad you asked.

The 2200 series (2232, 2248) need you to run NX-OS version 4.2 or later.

The actual version you are running on your Nexus 5000 is not capable to communicate with the 2232. Please consider an upgrade.

To decide on which version to go with, I recommend you to take a look at the release notes:

http://www.cisco.com/en/US/products/ps9670/prod_release_notes_list.html

Also here is the upgrade instructions, in the upgrade / downgrade guide:

http://www.cisco.com/en/US/docs/switches/datacenter/nexus5000/sw/upgrade/421_n2_1/Cisco_Nexus_5000_Series_NX-OS_Software_Upgrade_and_Downgrade_Guide__Release_421_chapter1.html

dhingst Wed, 03/02/2011 - 13:40

Hi Lucien,

I have a pair of new 5548's running 5.0(2)N2(1). I am trying to use Dell R810 servers with Qlogic QLE8150 CNA's. Under Windows 2008R2 I can't seem to pass a packet larger than 251 bytes through the switch from a 1Gb device to a 10Gb devices. Between 1Gb devices everything works fine. I have not tried between two 10G devices yet. From Windows a simple "ping 10.40.3.227 -l 251" works but using "-l 252" fails. This happens with multiple CNA's. The CNA itself appears to work since small pings work and the FC interface is up (but I haven't tried using any FC resources yet). Possibly related, Using the same model of server and CNA with VMware when the driver for the CNA is loaded the switch port will error out after 45 minutes or so with a "DCX-No ACK in 100 PDUs"  I'm scratching my head trying to figure out how to troubleshoot this so I can figure out if it is the switch, CNA, server or ??? Do you have any suggestions on things to look at before I call TAC? Since I don't have other cards I'm having a hard time pointing the finger at the right component. So far the 5548 makes a fine 1Gb switch but we haven't been able to do anything with 10Gb or FCoE.

Lucien Avramov Wed, 03/02/2011 - 21:15

What you observe is normal, the 5548 with the actual software revision actually does not support yet the speed 1 GB, even though you can configure it on the CLI. From version 5.0.3 that will be released in the next few days, the speed 1 GB will be supported and then I think your issue will go away. At this point you were able to configure 1 GB and have 1 GB to 1 GB work, however the issues start happening when you go from 1 GB to a 10 GB.

In the meantime, while you wait for 5.0.3 release, a  quick test for you to confirm this would be to remove the speed 1000 from each physical 5548 port and try 10 gb to 10 gb connection.

Lucien Avramov Thu, 03/03/2011 - 12:34

5.0(3)N1(1) just got released on cisco.com and supports 1 GB. Can you please upgrade? This should resolve your issue.

karthik.gopalak... Fri, 03/04/2011 - 11:47

Hello Lucien,

I have a question about Nexus 5548P Switches and the 1GB ports. Once I upgrade the switches, to 5.0(3)N1(1) - will I be able to direct connect a Cisco switch (maybe a 2960G) to the 5548 via SFP? Please let me know.

Also, a doubt regarding the Layer 3 upgrade to the 5548P, using the daughter card - When I am performing the initial configuration on the 5K - can I upgrade them to the 5.0.3.N1.1 and then setup vPC etc - so that adding the daughter card can be done one switch at a time (so that there will be no downtime)?

Thanks in advance!!


Lucien Avramov Fri, 03/04/2011 - 15:39

1. Yes if you upgrade to 5.0(3)N1(1) you will be able to use any port on the 5548 as 1 GB speed and connect for example a 2960G.

2. You can configure your switches and then upgrade to 5.0(3)N1(1), or first upgrade and then configure vPC. Remember that in the 5548P, you will need to remove the power supplies in the back place the card and replace the power supplies. This upgrade can be done in rack, you dont need to open the chassis.

Let me know if you have any other questions.

karthik.gopalak... Tue, 03/08/2011 - 10:03

Hi Lucien,

I am trying to upgrade the image on the 5548P to the 5.0(3)N1(1) code - I've downloaded the kickstart and system images, but when I run the "install all ...." command, I get an error - when I searches on some forums - it said the image could be bad. Any suggestions on why I am getting this?

NX5K-1#dir bootflash:

    4497974    Mar 08 16:03:46 2011  n5000-uk9-kickstart.5.0.3.N1.1.bin
  188747641    Mar 08 16:08:22 2011  n5000-uk9.5.0.3.N1.1.bin
NX5K-1(config)# install all kickstart bootflash:n5000-uk9-kickstart.5.0.3.N1.1.bin system bootflash:n5000-uk9.5.0.3.N1.1.bin
Verifying image bootflash:/n5000-uk9-kickstart.5.0.3.N1.1.bin for boot variable "kickstart".
[#                   ]   0% -- FAIL. Return code 0x40450009 (cannot read image info).
Pre-Upgrade check failed. Return code 0x40930011 (Image verification failed).
NX5K-1(config)# sh install all status
This is the log of last installation.
Verifying image bootflash:/n5000-uk9-kickstart.5.0.3.N1.1.bin for boot variable "kickstart".
FAIL. Return code 0x40450009 (cannot read image info).
Pre-Upgrade check failed. Return code 0x40930011 (Image verification failed).
Any help would be appreciated!
Lucien Avramov Tue, 03/08/2011 - 10:39

The size of your kickstart image does not seems right, it should be: 25148416 bytes, the MD5 checksum for it is: e5c733cdd70feac0ee746924d0372709

Your system image has the correct size.

I advise you to download again the kickstart from our website and compare the size, MD5 if need be and then run install all again.

karthik.gopalak... Tue, 03/08/2011 - 11:41

Yes sir!

I downloaded the kickstart again from CCO - and it worked. Thanks for your help!

thorton.stevet Fri, 03/04/2011 - 11:07

Hi Lucien,

I have 5020s and a 5548, can I use the ports with a speed of 1 GB? Thanks Steve.

Lucien Avramov Fri, 03/04/2011 - 15:40

Yes you can, here are the caveats:

1. On the 5520s, you can use the first 16 ports as a speed 1 GB (port e1/1 to e1/16).

2. On the 5548, you can use any port with a speed of 1 GB, however you need run the latest NX-OS version: 5.0(3)N1(1). That code was released yesterday on cisco.com, so you may download it.

mohsin.khan@tel... Sun, 03/06/2011 - 21:56

Hi Lucien,

Thanks for being here with us for our help.

Nexus 5K can support only 12 N2K fiber extenders at a time. That is out of 20 ports, only 12 ports can be used to connect to N2Ks. Is this a true statement, if yes then whats the purpose of remaining 8 ports on N5k?

How many (and which) ports on 5520 can be used for VPC with N2K?

As per my understanding, VPC is configured on the downlink ports, plz confirm.

Nexus 2K fiber extender modules cuurently do not support link aggregation. Why do we have it like this, and is there any roadmap for availability fo this feature?

If it is required to dual-home the downstream switch to N2K, Flex Link feature should be used. This way spanning-tree can be avoided. Can N2K be used to connect the downstream switch in this way, and would there be more cons than the pros?

regards,

Mohsin

Lucien Avramov Mon, 03/07/2011 - 13:23

Great questions.

As you say, the 5000, can support up to 12 FEX, and the 5500 up to 16 as per:

http://www.cisco.com/en/US/docs/switches/datacenter/nexus5000/sw/configuration_limits/limits_503a/nexus5k_config_limits_503_chapter1.html

Keep in mind that the 2148,2248 have 4 10 GE ports, the 2232 have 8 10 GE ports and the 2224 2 10 GE ports, so depending on model and the bandwitdh you will allocate to the FEX you will use between 1 to 8 10 GE ports for each FEX.

The ports on the Nexus 5000 series can be used for multiple type of connections: connection to other switches (Nexus 7K, Catalyst switches etc..), connection to a FEX, connection to 10 GB servers.

The FEX do support LACP. There are two scenarios:

-2148 series: a port-channel with LACP accross 2 2148 works when you configure vPC. Port-channel within 2 physical ports on the same 2148 will not work.
-2200 series: LACP will work with either vPC or multiple connections on the same FEX.

The N2K can have a switch connected downstream, however you need to be very careful with not introducing a loop in your network since there is no spanning-tree running on the FEX ports. To get your link between a FEX and a switch to come up, you will have to enable spanning-tree bpdufilter, so the link doesnt error disable.

xjullien84 Tue, 03/08/2011 - 08:21

Hi Lucien,

I have 2 Nexus 5020 running NX-OS 5.0(2)N1(1) and 1 N2248..

The VPC architecture is: FEX Dual-Connected to N5K switches

VPC Peer link between both N5K is UP.

N5K ports connected to the N2K are in “"fex-fabric DOWN” reason= SDP timeout/SFP Mismatch

I’ve tried to configure the same ports in “access mode” and the port comes UP.

As soon as I configure them in “fex-fabric mode”, the ports go down.

Here is my N5K configuration:

int port-channel 102

description FEX-102

switchport mode fex-fabric

fex associate 102

vpc 102

int e1/20

description FEX-102

switchport mode fex-fabric

fex associate 102

channel-group 102

Can you help ?

Thanks ;-)

Xavière

Lucien Avramov Tue, 03/08/2011 - 10:43

Thanks for your question. I assume you have only e1/20 connected on each N5K to this fex.

Can you make sure of the following, and feel free to attach the outputs back:

-show interface e1/20 transceiver details

I suspect you are not using matching SFPs on each N5K. You can use any supported SFP from the matrix, however we need them to match on each N5K when going to the same FEX.

http://www.cisco.com/en/US/docs/interfaces_modules/transceiver_modules/compatibility/matrix/OL_6974.html#wp55411

xjullien84 Wed, 03/09/2011 - 06:01

Thx a lot !

SFP didn't match :-) It works now !

Xavière

apmorgan Tue, 03/08/2011 - 13:21

Hello,

I am trying to extend the fabric from a 5020 to a 2248TP and 2232PP.

I have the fet-10g but my 5020 doesn't like it. Link light is orange on the 5020 port and off on the 2000 switches.


Eth1/17       --                 sfpInvali 1         full    10G     10g
Eth1/18       --                 sfpInvali 1         full    10G     10g

Ethernet1/17
    transceiver is present
    type is Fabric Extender Transceiver
    name is CISCO-FINISAR
    part number is FTLX8570D3BCL-C1
    revision is A
    serial number is ......

    nominal bitrate is 10300 MBit/sec
    Link length supported for 50/125um fiber is 20 m
    Link length supported for 50/125um fiber is 100 m
    Link length supported for 62.5/125um fiber is 10 m
    cisco id is --
    cisco extended id number is 4

           Transceiver Detail Diagnostics Information (internal calibration)
  ----------------------------------------------------------------------------
                Current              Alarms                  Warnings
                Measurement     High        Low         High          Low
  ----------------------------------------------------------------------------
  Temperature   35.61 C        75.00 C      5.00 C     70.00 C       10.00 C
  Voltage        3.27 V         3.63 V      2.97 V      3.46 V        3.13 V
  Current        0.40 mA  --   11.80 mA     4.00 mA    10.80 mA       5.00 mA
  Tx Power      -14.81 dBm --    2.69 dBm  -11.30 dBm   -1.30 dBm     -7.30 dBm

  Rx Power        N/A     --    2.99 dBm  -13.97 dBm   -1.00 dBm     -9.91 dBm
  ----------------------------------------------------------------------------
  Note: ++  high-alarm; +  high-warning; --  low-alarm; -  low-warning

FET-10G is documented to be support on the Nexus 5020 with the 2000 switches. I tried with NX-OS of 4.2 and latest 5.0.3 neither work.

Any ideas what is wrong?

Thanks

A.

Lucien Avramov Tue, 03/08/2011 - 17:17

Have you configured the port as switchport mode fex-fabric? The FET-10G only works for this mode.

apmorgan Wed, 03/09/2011 - 10:55

Thanks,

switchport mode fex-fabric

did the trick.

Lucien Avramov Wed, 03/09/2011 - 07:40

You are correct  SFP-10G-ER is not supported at this time on the 5500. It is being evaluated for support in future code releases, no more details to share as far as when at this time.

mohammedah Wed, 03/09/2011 - 10:17

Hello Lucien,

I just want to know if ERSPAN is supported on the Nexus N5K-C5020P-BF which is running 5.0(2)N1(1). It seems the commands are there to set up ERSPAN source however the parameters like destination IP and source interfaces are not saved in  and the ERSPAN indicates as being down.

Here is sample output:

D45-N5K1-25# sh monitor session 1
   session 1
---------------
type              : erspan-source
state             : down (No src/dst configured)
ERSPAN ID         : 100
ERSPAN TTL        : 3
ERSPAN IP Prec.   : 0
ERSPAN DSCP       : 0
ERSPAN MTU        : 1500
source intf       :
    rx            :
    tx            :
    both          :
    rx            :
destination IP    :

D45-N5K1-25#

Lucien Avramov Wed, 03/09/2011 - 10:44

Thanks for asking. The issue is reported in CSCtn84269, you can view this in our bug tool kit page:

http://tools.cisco.com/Support/BugToolKit/action.do?hdnAction=searchBugs

Basically, in 5.0.2 code this shows in the parser, however the feature is still not supported. In 5.0.3 and later code, this has been removed from the parser and is not an option.

Rest assured, the feature ERSPAN is in the works but is just not yet released.

paveldimow Wed, 03/09/2011 - 13:33

Hi Lucien,

is it possible to have FEX 2148 connected to two Nexus 5k for redundancy? If I look at FEX like a linecard I think it's not possible but

can you confirm this please?

Tnx

jecheand@pe.ibm.com Wed, 03/09/2011 - 19:56

Hi lucien,

i had received my two first Nexus 5000 but one of them arrived without an NX-OS image and in Loader> prompt.

I tried to load by TFTP the kickstart image but the system gave me the following error message:

loader> boot tftp://172.16.1.3/n5000-uk9-kickstart.4.2.1.N2.1a.bin

Address: 172.16.1.2

Netmask: 255.255.255.0

Server: 172.16.1.3

Gateway: 172.16.1.1

Booting: /n5000-uk9-kickstart.4.2.1.N2.1a.bin console=ttyS0,9600n8nn quiet load

er_ver="pr-1.3"....ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

ALERT: got a fragmented packet - reconfigure your server

TFTP error 0 (Unknown request code.)

Error 15: File not found

I tried this process with 2 different TFTP software 3cDaemon and Solarwinds but the problem persists.

Do you have any advice on these.

Thanks,

Pepe

jecheand@pe.ibm.com Thu, 03/10/2011 - 06:32

Hi Lucien, it works! but we have to enable the parameter PXE Compatibility in the Advanced TFTP Options to work.

Thank you very much.

Pepe

s.leyer Thu, 03/10/2011 - 03:36

Hi Lucien,

LACP is supported on FEX with N5K. Recently we have attached 2 2248 to a 7010
(running 5.1.2) and it's not allowed to build an endhost port-channel (no vPC)
on FEX ports. Message: "channel-group not supported for interface type".
feature lacp is enabled.

I cannot find any documentation about this issue.
Is it a bug or really not supported ?

Thanks
Silvia

Lucien Avramov Thu, 03/10/2011 - 10:14

You are correct, this is not yet supported when the FEX is attached to the N7K. However this is just a software limitation at this time on the N7K, and future release of code will provide you this functionality.

In the meantime you can have a vPC and connect this server across two fabric extenders.

dtran@behr.com Thu, 03/10/2011 - 14:19

Hi Lucien and thanks in advance for your time !!!

I am looking into a N7K and N2K deployment scenario and have a few questions.

1 - I understand as of today you can only run port-channel between an N2K and a single N7K (vPC is not supported). Is this correct ?

2 - Will vPC be supported between N2K and N7K in the near future ? Do you have a timeline when the next NX-OS release will be out that supports vPC between N2K and N7K ?

3 - With the N2K running port-channel to N7K. When you upgrade the software on the N7K, the N2K will automatically pull its new software image from the N7K then reboot correct ? if so, does this happen automatically after the upgrade is complete on the N7K ? is there any way to schedule the N2K to run its upgrade at a later time to mininize production interruption since the N2K upgrade requires a reboot ?

4 - Do you foresee any issues / concerns in this deployment scenario ? I would love to be able to vPC between N2K and N7K.

Thanks again Lucient !!! Looking forward to hear back from you !!! very much appreciate your help !!!

D.

Lucien Avramov Thu, 03/10/2011 - 15:21

Actually there are some good news here: you can run vPC on N7K for FEX starting 5.1 code.

The FEX downloads it's code from the N7K.

The upgrade of the OS of the 7K will push the new code onto the FEX and reload the FEX at the sametime the 7K is reloaded, you can not avoid that as far as I know.  So feel free to upgrade to the 5.1 release on N7K and use vPC on your FEX.

nils.neumann Thu, 03/10/2011 - 23:49

Hi Lucien,

this is not a nice software limitation ... port-channels are very important in a datacenter.

Is the limitation solved in 5.1(3) or 5.2(1) ???

Lucien Avramov Fri, 03/11/2011 - 11:59

From 5.2 you can have port-channel to the same fex or over vPC

mario-leitao Thu, 03/10/2011 - 12:22

Hi

When will the L3 daughter card be available for the nexus 5K and what will be

the partnumber for it?

Lucien Avramov Thu, 03/10/2011 - 15:17

The card is available for ordering! The part number for the daughter card, for 5548 is :N55-D160L3.

If you have a 5596, you can order an extansion module, the part number for the L3 module for the 5596 is: N55-M160L3

dtran@behr.com Thu, 03/10/2011 - 16:18

Hi Lucien,

That's excellent news !!! I understand a lot of customers have used vPC from N5K to the FEX. But vPC from the N7K to the FEX is new. Have you seen customers use vPC between N7K and FEX ? Do you foresee any issues / concerns with vPC between N7K and FEX ?

Thanks again Lucien !!! very much appreciate your help !!!

D.

Lucien Avramov Thu, 03/10/2011 - 16:21

Sorry, I was maybe a bit confusing, you can do vPC between the FEX and the end host (servers) from 5.1.

vPC between 7k and FEX is not supported. Do you see a need for it? You can have dual supervisor on the N7K, so there goes your high availability.

dtran@behr.com Thu, 03/10/2011 - 16:41

Yes, I would like to be able to run vPC from the N7K to the FEX.  I am actually on a project working on replacing my core 6509 with a pair of N7K switches and I am looking at the N2K for Top-of-Rack. Please clearify the questions below for me !!! I don't see the need to have the N5K at this time for my environment. Thanks Lucien !!!!!!!

1 - I understand as of today you can only run port-channel between an  N2K and a single N7K (vPC is not supported). Is this correct ?

2 -  Will vPC be supported between N2K and N7K in the near future ? Do you  have a timeline when the next NX-OS release will be out that supports  vPC between N2K and N7K ?

3 - With the N2K running port-channel to  N7K. When you upgrade the software on the N7K, the N2K will  automatically pull its new software image from the N7K then reboot  correct ? if so, does this happen automatically after the upgrade is  complete on the N7K ? is there any way to schedule the N2K to run its  upgrade at a later time to mininize production interruption since the  N2K upgrade requires a reboot ?

4 - Do you foresee any issues / concerns in this deployment scenario ? I would love to be able to vPC between N2K and N7K.

Lucien Avramov Thu, 03/10/2011 - 16:52

1. Yes

2. Not that I am aware of, but I can speak for what is released at this point. Your sales account team, will help you as far as roadmap. Again why would you even need this feature on an N7K? You just need to have 2 supervisors already on that chassis and that will take care of the redundancy.

3. Yes it does, it's automatic, there is no way to postpone this to later time, the upgrade happens along with the N7K. Again to minimize the downtime the FEX reboots as the same time the 8k does.

4. I dont see issues

dtran@behr.com Thu, 03/10/2011 - 23:24

Thanks Lucien !!!! I see your point in regard to having dual Supervisors in the N7K and that's what I will have (dual Sup's in each N7K) but I like the idea of having the extra layer of redundancy by having the N2K connecting to each N7K via vPC in case one of links fails or one of the N7K's completely dies. Please share your thoughts on this !!!

You've mentioned vPC can be done between FEX and end host (servers). Let's say that I have two N7K's (N7K-1 and N7K-2) connected via vPC and the FEX is connected to N7K-1 via port-channel. Can I vPC to an end-host with one link connects to the FEX and one link connects N7K-2 ?

Also similar question but in a different scenario. In this scenario I also 2 N7K's (N7K-1 and N7K-2) connected via vPC with a FEX connects to each N7K via port-channel, so one FEX connects to N7K-1 and one FEX connects to N7K-2. Can I vPC to an end-host with one link going to each FEX ?

Thank you very much Lucien !!!! You have been very helpful !!! and I very much appreciate your help !!!!!!! please let me know if my questions aren't making sense.

D.

Lucien Avramov Fri, 03/11/2011 - 11:57

You can have multiple line cards on 1 N7K, and connect multiple uplinks between a FEX and 7k. Added the dual supervisors, you have a total redundancy as far as the 7k is concerned. You can also have multiple FEXs and dual home server to 2 FEX in case of a FEX failure. Your FEXs needed to be connected on the same N7K, for a server to be dual homed to 2 FEX.

r.westman Fri, 03/11/2011 - 06:46

Hi,

We have a ESX-machine connected with a port-channel (LACP) to two 2248, each 2248 singled attached to a separate 5548. When we power down both 5548 and then turn on only one of the 5548, the ESX machine never comes online. The 5548 takes the primary role after reboot, the VPC goes up, the interface connected to the ESX machine goes active. And the LACP negotiation seems to be ok. But the ESX machine never answers any ping until the other 5548 is turned on and the peer-link is formed again. What can be wrong?

Regards

Interface Ethernet102/1/13 is up

  Channel group is 13 port channel is Po13

  PDUs sent: 63

  PDUs rcvd: 58

  Markers sent: 0

  Markers rcvd: 0

  Marker response sent: 0

  Marker response rcvd: 0

  Unknown packets rcvd: 0

  Illegal packets rcvd: 0

Lag Id: [ [(7f9b, 0-23-4-ee-be-14, 800d, 8000, 230d), (ffff, 0-1c-c4-d8-fd-66, 11, ff, 1)] ]

Operational as aggregated link since Fri Mar 11 14:03:38 2011

Local Port: Eth102/1/13   MAC Address= 0-5-73-c3-4b-fc

  System Identifier=0x8000,0-5-73-c3-4b-fc

  Port Identifier=0x8000,0x230d

  Operational key=32781

  LACP_Activity=active

  LACP_Timeout=Long Timeout (30s)

  Synchronization=IN_SYNC

  Collecting=true

  Distributing=true

Actor Admin State=(Ac-1:To-1:Ag-1:Sy-0:Co-0:Di-0:De-0:Ex-0)

Actor Oper State=(Ac-1:To-0:Ag-1:Sy-1:Co-1:Di-1:De-0:Ex-0)

Neighbor: 0/1

  MAC Address= 0-1c-c4-d8-fd-66

  System Identifier=0xffff,0-1c-c4-d8-fd-66

  Port Identifier=0xff,0x1

  Operational key=17

  LACP_Activity=active

  LACP_Timeout=Long Timeout (30s)

  Synchronization=IN_SYNC

  Collecting=true

  Distributing=true

Partner Admin State=(Ac-0:To-1:Ag-0:Sy-0:Co-0:Di-0:De-0:Ex-0)

Partner Oper State=(Ac-1:To-0:Ag-1:Sy-1:Co-1:Di-1:De-0:Ex-0)

Lucien Avramov Fri, 03/11/2011 - 12:05

ESX nor ESXi support LACP. You can do a port-channel without lacp (channel-group mode on).

If you need LACP support you can use the Nexus 1000v on ESX.

lnxnut.certeon Fri, 03/11/2011 - 08:21

Hello Lucien,

I have been working with ESX and ESXi for a couple of years and am now just getting into using the nexxus.

Is there a "trial" period that I can download and use it before I have to come up with a PO?

Thank you,

Rick

Lucien Avramov Fri, 03/11/2011 - 12:06

Sure, you can try it for a 60 days period and download it from our website.

tbleise Fri, 03/11/2011 - 00:27

Hello Lucien,

I try to find details about the ingress/egress buffer size and QoS characteristics of the Nexus 2248. While there is a nice document about the Qos of Nexus 5000 on

http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9670/ps11215/white_paper_c11-622479.html

I could not find something similar for the Nexus 2000 (in particular 2248).

My questions are

1. what is the ingress and egress buffer size for each port?

2. does the Nexus 2248 implement VOQ between the host ports and the fabric ports?

    If not, what buffering scheme is deployed?

    If yes, how many ingress / egress queues are there? Can the buffer allocation be configured per system class, similar to Nexus 5000?

3. Do all the QoS commands described for the Nexus 5548 (like policy map attachment, buffer allocation per class etc.) only apply to "directly" connected Nexus 5000 ports or does it also apply to Nexus 2000 Ports?

   If yes, are there limitations? May depending on the Nexus 2000 model attached?

   If no, which command is supported and which is not?

Thanks in advance for your help!

Best regards,

   Thorsten

Lucien Avramov Fri, 03/11/2011 - 14:37

Great questions. The FEX depend on the buffer of the 5k.

The 2148 has better buffer management from 4.2.1 code of NX-OS.

There is about up to 320/375 kb per 4 ports as buffer (shared buffer)

There is a cli command to use so one port can absorb more shared buffer from the 4 ports to absorb burst, that would be useful in isci for example for 2148:

N5k(config)# system qos

N5k(config-sys-qos)# no fex queue-limit

The default threshold works fine on 2248 and 2232.

You configure the QoS for 2248 from the N5K via the service
-policy.

Actions

Login or Register to take actions

This Discussion

Posted February 25, 2011 at 1:10 PM
Stats:
Replies:49 Avg. Rating:5
Views:45401 Votes:2
Shares:0

Related Content

Discussions Leaderboard