Cisco Switching/Routing :: 5596 LACP Implementation With HP Servers
Apr 17, 2012
I need to implement LACP HP servers mostly DL 380 g7 with Intel based dual port with two types of Cisco equipment first scenario server connected to 3750x stack of 4 switch's .second scenario same server type connected to two Cisco Nexsus 5596 . My question regarding two type of connection.Is it possible to do active active ?Would it give fault tolerance ?With HP LACP implementation is there known issue or should i expect latency with such configuration?What is the maximal lag- channel group that is possible per type?
I have two N5K (5020) switches with NX-OS - 5.0(3)N2(1). These two switches form VPC domain: peer-link = 2*10Gb ports (1/17-18) and peer-keepalive link over managements ports.Also I have two HP servers with two 10 Gb ports on each server.Each server conected by one link to each N5K switch (1/9-10). N5K downlinks configured as access ports with LACP Active mode.There is only one VLAN (1).When "no shut" command entered on N5K access ports - ports going in "not connected" status, begin flap and then going in "linkFlapErrDisabled" state.In attach - "sh run" from N5K.
I have a network with four 6509s in a ring with 10Gb links. Two adjacent switches are at the home office, the other two at the DR site. The switches at each location are physically similar to each other with respect to what blades are in them. We went through an upgrade from SUP-720's to VS-SUP-720's recently, only at the DR site - basically a practice, with the home office conversion hopefully taking place next weekend.
We initially just brought up the two chassis separately, in non-VSS formation (stand-alone). So far, so good - everything was connected, all traffic was passiing, all links were up, everything was reachable: EVERYTHING worked. Then we made the conversion: step-by-step from the cisco.com page; create a virtual domain, make one switch switch 1, the other switch 2, create differently numbered port-channels on each 6509, add the SUP 10Gb links to the port-channel, do the conversion.
Here's where the trouble started. First of all, the two 10Gb links back to home office created a spanning-tree loop and we had to shut down one of the links. (Is there something that needs to be configured on those links to turn spanning tree on? Does VSS conversion turn stp off?) Secondly, though it worked while in stand-alone mode, the copper blade in the standby 6509 stopped passing traffic - it would take config, the links would come up, but you could not ping across those links. Interestingly enough, there was an access switch with links to each of the copper blades, and having them both up also caused a spanning-tree loop. adding a new port-channel and putting both links in it did nothing to alleviate the loop. This leads me to believe that stp is not working properly. I reiterate, that even though the loop occurred, nothing else plugged into that blade was pingable.
I'm looking to try and implement ipv6 HSRP on a series of IOS-XR Routers running 4.2.1 following on from successfully setting up IPv6 HSRP on a few cat6509s on VLAN Interfaces in other parts of the network. I have entered the "router hsrp" configuration menu and gone into the interface in question that I'm looking to setup with IPv6 HSRP. Unfortunately, there version 2 or address-family ipv6 commands are not available.
We are going to be expanding our Shoretel phone system in our HQ and I need to get QoS configured correctly. All of our offices are connected via MPLS and I need to make sure that we are sending QoS tagged traffic to our provider. The phones are tagged by the director, but there is other traffic for call control that needs to be tagged. I don't have access to our CPE router as it is managed by Sprint. The Sprint router is connected to our internal network. We have our data network running on (4) 3750x switches running 12.2(55) with IP feature set.
The problem is that many of these commands don't work on the 3750 (priority, bandwidth, match protocol, etc...) and the configuration assumes you are applying this to an outbound queue which is not supported on the 3750. I think I have to do this with policing, but I'm not sure what interfaces need to have this applied.
I am planning an implementation of VSS on our two 6509 switches, and would like some feedback on things to look out for, and any issues encountered by others that have done this already.
We have the 10Gig port installed on the Management blades, but not configured yet.
Main questions would be:
1: What kind of "down time" am I looking at for the migration? (Reboots, configuration reloads, etc.)
2: I will be saving the configurations on both devices before-hand, but how does the VSS migration "merge" the configurations of both devices?
3: L2 VLANS - we have some on one switch, others on the second switch. Will these be combined, or would this be a manual process?
Any other things of note that I should know about before planning this migration?
I have 2 Nexus 5596UPs with a layer 3 cards that are exhibiting some very peculiar behavior. The systems are running 5.1(3)N1(1).I have configured 2 VRF contexts each running their own OSPF process. There is a static gateway of last resort configured on each VRF, which is to an upstream pair of 5585X's in Active/Active. Each OSPF process has the "default-information originate always" command configured, however, backbone neighbors are not recieving a gateway of last resort from the 5596UPs. The applicable configurations are show below. All other routing information is passing correctly between devices in the network. This network is not production, it is a proof of concept for a larger implementation.
- Second question is about EIGRP, when I configure EIGRP on the main switch that is AVG with the following commands, will I also have to run the same commands on the second 4948 E too?
We have setup a pair of Nexus 5596 L3 switches with 2 x 10Gbps LACP links between them to act as the vpc peer link. We also have another 2 x 10Gbps LACP links between the 5596 switches to carry non VPC VLANs, this is required to provide EIGRP routing between the switches and an upstream router.I have read that it is possible to setup the vpc keep-alive link over an SVI instead of the management interfaces. Is it ok to run the keep-alive SVI over the second LACP non VPC VLAN trunk or is it recommened to keep this seperate?
I am looking to see if Nexus 5596UP & Nexus 2248TP GE compatible with SFP-10G-SR? The reason is because a consultant was hired on to "design" the network layout and has decided to purchase Cisco SFP+ Copper Twinax Cables which have a 10M limit. A small handful of the Data Center racks are 10-15M away... just out of reach of the Twinax. I would prefer NOT to move the LAN row so that it is more centered in the room. Can I use the SFP-10G-SR to connect the 2 switches (5596 & 2248) together? This SFP has a 26M reach on standard 10gig fiber, the small cost increase per connection is of no concern.
i have: two nexus 5596 connected each other the mgmt0 is NOT in use SVI for keepalives with IP address and /30 netmask vpc-keepalives running over fiber in e1/1. this works well uplinks to datacenter distribution switch (Cat 6500 VSS) over fiber on port-channel 1 (e1/2 and e1/10), also carrying the management VLAN (vlan 14). SVI with an IP address for management purposes
I can't get this to work. i can ping my whole network from the nexus, but not the nexus from my network. also pinging inside the mgmt vlan is not possible.
we have configured VpC between two Nexus 5596, for Vpc-Keep-alive-link we configured L3 interface with 1G (GLC-T) ,it shows the below status message "L3 not Ready" with interface LED glows in Yellow in color. is this a physical layer problem
Ethernet VLAN Type Mode Status Reason Speed Po Ch Eth1/17 -- eth routed down L3 not ready 1000(D) -- Eth1/18 1 eth access down SFP not inserted 10G(D) -- Eth1/19 1 eth access down SFP not inserted 10G(D) -- Eth1/20 1 eth access down SFP not inserted 10G(D) -- Eth1/21 1 eth access down SFP not inserted 10G(D) --
For a simple 4x sites design for backup consolidation (3x sites connected to one central site), does the design in the attached JPEG make sense and is the configuration achievable using the specified parts?
Brief Explanation: - Each site is to have servers with 2x 10Gb/s Ethernet (both teamed/bonded/etc) and one NIC on each of the 2x switches in the site (if possible). Each site will also have 2x Cisco Nexus 5596UP Switch (though the P version may be used instead of the UP). - Two of the sites are within 10KM of the central site - The last site is about 35KM from the central site - It is required to have a minimum bandwidth of 10Gb/s between any site and the central site - I have specified 10Gb/s LW SFP+ and 10Gb/s LW SFP+ (for the 2x 10KM and 1x 35KM sites) - are there any special licenses needed? - as asked before, can such a "simple" design work as-is? - what sort of single-mode fiber would fit the bill (especially for the 40KM link) - 9nm?
I m planning to implement VSS in core but want some inputs on IOS as i have FWSM as a service module Core :- Ii am running 12.2(33)SXH2a on my Core 6509 and i checkd cisco sites and Fwsm release notes but it states only I-Train of IOS while mine is H-Train so can I directly upgrade to I-Train or I was thinking of SXH8b IOS.
We have HSRP between NexusA and NexusB with access layer switches connecting to the core using VPC, We are trying to setup a VAM server Voice recording for Siemens phones. We need to span all voice vlan and point it to the VAM server the VAM server connects to a 3750 Stack considering the amount of traffic multiple span session can generate I plan to move the server to the Nexus directly and run a Local Span Session.
1- As we have two Nexus running HSRP and VAM server only connects physically to one NexusA (I can run local span on that nexusA) the Second NexusB is not directly connected to the VAM server I plan to run ERSPAN so if this is the best design and which path will the span traffic take from Nexus B to NexusA will it go through the access layer switches depending on the vlans allowed on the uplinks or will it go through the 20 Gig uplink between the two Nexus allowing all vlans (VPN peer links) ? WE have approximately 10 voice vlans, Do we an example config for ERSPAN session where the source are vlans (As I am for fimilliar with RSPAN) ?
It looks like the deny statement is not working as I can see all routes I am redistributing. I even did a deny on a specific route and I still see it in the routing table on another router in the autonomous system.The same below works fine on IOS platform. [code]
I have a pair of 5596 running in a vPC with Nexus 2248 connected to each N5596. When I do the command "show fex" I get the following output on the 2nd 5596
Number Description State Model Serial ------------------------------------------------------------------------ 101 FEX101 AA Version Mismatch N2K-C2248TP-E-1GE SSI16390705 102 FEX102 AA Version Mismatch N2K-C2248TP-E-1GE SSI163704AD 122 FEX122 Online N2K-C2232PP-10GE SSI16370195
I'm running version 5.1(3)N1(1) on both of the 5K's. I have looked through all the configuration and I am not understanding why I am getting this error. I have tried to look it up on [URL], but not having a ton of luck.
I have 2 datacenters running same equipement (two Nexu 5596 with FEX).I just took a look at the log just to see if everything is ok and I saw that I have the same error message (a lot of times) at both location :
%SYSMGR-FEX100-5-HEARTBEAT_LOSS: Service "satctrl" heartbeat loss 2 ,max 7,I though it was a problem with my peerklink-keepalive connection but I see the word FEX ....so i'm not sure...
Note that at both locations, my Nexus are connected back to back through the management port using transceivers. So it's a copper cable from the first nexus, going into a transceiver, going to another transceiver in fiber and then back to copper to the other nexus.
I've got LACP-enabled port-channels between a Cisco 3750 stack and a few different switches (some Cisco 3750s and some Juniper EX2200s). The Ciscos are all sending slow LACP updates, the Junipers are sending fast LACP updates (but the Cisco they connect with is responding with slow LACP updates).
I have a couple of questions:
1) what are the pros and cons of slow vs fast updates? my research has led me to the conclusion that fast updates are better for network resiliency as long as you have plenty of bandwidth overhead (which I do at the moment). is there anything to add to this conclusion?
2) is there any way to configure the Cisco 3750s for fast updates?
We're trying to configure our Cisco 4507 (Supervisor Engine IV) to allow a new Dell server with a pair of Broadcom 5708 GigE NIC's to aggregate its NIC's to give us a 2gbps link to the switch.
So far we seem to have got the team and LACP up and enabled, but the adaptor that the Broadcom Admin Util creates for the team is only showing a 1gbps connection where I would have expected it to show as 2gbps.
The individual NICs show as connected at 1gbps. We're not Cisco experts so are struggling on how to get the 2 NICs to aggregate.
On the server side we've done nothing other than create a team using 802.3ad LINk Aggregation using LACP.
This is what I think the relevent output from "sho conf" is, more available if needed.
version 12.2 boot system flash bootflash:cat4000-i9s-mz.122-18.EW1.bin ! interface Port-channel2
i'm desperately trying to get LACP working over a dot1q Tunnel. The "Service Provider" Switches are two 4506-E Switches with SUP7-E connected via a 10G Link, running on cat4500e-universalk9.SPA.03.03.00.SG.151-1.SG
sample config:
dot1q tag vlan native interface GigabitEthernet3/1 switchport access vlan 2001
I am trying to configure a 6509 as the passive receiver from a Dell Force10 10Ge switch with 2 sfp to 2 gig ports on our 6509 switch, I see LACP is up on both sides but cannot pass traffic, I have only 2 vlans that will carry across the aggregate link from our vmware boxes, this is just a temp until I get a 10ge in our 6509 chassis.
I have conneccted a server with LACP on Nexus extender. I am starting different file copy from diffferent sources to this server, it does not load balance.
We have 2x WS-C3560X-48 with the 10Gb SFP C3KX-NM-10G module.I want to use the 10Gb SFP (with a redundant 1Gb link) between each of the switches.Below is the configuration what I think I should use. Is this correct?
SWITCH1 int TE1/1 switchport trunk encapsulation dot1q switchport mode trunk channel-group 1 mode active
[code]....
We also want to utilise Link Agrrigation for the servers that attach to the switch. Would this config be correct?
port-channel load-balance src-ip int range x switchport mode access switchport access vlan 1 (They are all on the default vlan) spanning-tree portfast channel-group x mode active
(Then I would configure the LACP config on the server) Is there anything I am missing?
I am deploying a pair of Nexus 5596's with 3750 POE switches in the closets. I'm looking for a best practice as how to configure the Nexus 5596 to support proper QoS for EF at the core.
I currently have Nexus 5596 pair with VPC peer link Po1 between them. My goal is to connect our new Nexus 7Ks to the 5K's using Fabric Path. My question is during this inital setup with the 7K's. Can I use the same port channel number on the 7K's as I did the 5K's? Is the port channel locally significat?
i want to remove the sync-profile on each of two synchronized Nexus 5596UP without loosing the config stored in Switch-Profile. That means without connectivity interruption and re-configuration of interfaces in "conf t mode", for example. Since NX-OS Release 5.2(1)N1(1) there is a new command
switch(config-sync)# no switch-profile abc profile-onlyprofile-only—Deletes the switch profile without the local configuration.
