Dear Members, - Page 2

hemadri37 · ‎02-14-2017

Dear All,

I am facing an issue of continuous port flap.

There is one NETAPP FAS2020 2U connected to the cisco switches.

Port 1/8 of cisco switch A is connected to EOA in module A and port 1/8 of cisco switch B is connected to EOB in module A.

port 1/8 in both cisco switches is flapping continuosly. logs and existing configuration is attached.

please advise.

Attachment:

netapp.docx

These ports in both cisco switches are flapping continuosly.

2716783: Feb 1 08:22:59: %DTP-5-NONTRUNKPORTON: Port Gi1/8 has become non-trunk
2716784: Feb 1 08:23:00: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet1/8, changed state to down
2716785: Feb 1 08:23:01: %LINK-3-UPDOWN: Interface GigabitEthernet1/8, changed state to down
2716786: Feb 1 08:23:03: %LINK-3-UPDOWN: Interface GigabitEthernet1/8, changed state to up
2716787: Feb 1 08:23:05: %DTP-5-TRUNKPORTON: Port Gi1/8 has become dot1q trunk
2716788: Feb 1 08:23:05: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet1/8, changed state to up
2716789: Feb 1 08:23:07: %DTP-5-NONTRUNKPORTON: Port Gi1/8 has become non-trunk
2716790: Feb 1 08:23:08: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet1/8, changed state to down
2716791: Feb 1 08:23:09: %LINK-3-UPDOWN: Interface GigabitEthernet1/8, changed state to down
2716792: Feb 1 08:23:12: %LINK-3-UPDOWN: Interface GigabitEthernet1/8, changed state to up
2716793: Feb 1 08:23:14: %DTP-5-TRUNKPORTON: Port Gi1/8 has become dot1q trunk
2716794: Feb 1 08:23:14: %DTP-5-NONTRUNKPORTON: Port Gi1/8 has become non-trunk
2716795: Feb 1 08:23:16: %LINK-3-UPDOWN: Interface GigabitEthernet1/8, changed state to down
2716796: Feb 1 08:23:19: %LINK-3-UPDOWN: Interface GigabitEthernet1/8, changed state to up
2716797: Feb 1 08:23:21: %DTP-5-TRUNKPORTON: Port Gi1/8 has become dot1q trunk
2716798: Feb 1 08:23:21: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet1/8, changed state to up
2716799: Feb 1 08:23:50: %SPANTREE-5-TOPOTRAP: Topology Change Trap for vlan 200
2716800: Feb 1 08:24:10: %DTP-5-NONTRUNKPORTON: Port Gi1/8 has become non-trunk

pwwiddicombe · ‎03-07-2017

I've never configured a NetApp before, although I've configured the network side of the connection.

Based on what I see above, it looks like vif0 comprises 2 physical interfaces, one of which is down. From earlier posts these are connected to 2 DIFFERENT switches for failover purposes. You currently have e0b down to prevent flapping ?

It looks like vif0 is in load-balancing mode between the 2 interfaces (multi_mode), which is good for performance, as it load balances across both interfaces (giving you effectively 2 Gig throughput). However, this only works if the Cisco end knows about it, and has it configured as an etherchannel, which you don't. Hence the flapping when both links are up.

I believe what you want is vif-type single_mode to accomplish what you want. However, you might want to either google it, or check with either vendor or integrator for the exact syntax.

https://library.netapp.com/ecmdocs/ECMP1511538/html/man1/na_vif.1.html and look for the HA section. The example is a bit more complex - it has a 2-link bundle to one switch; another 2-link bundle to a second switch. Then the 2 virtuals are paired into an HA bundle where only one is active at any given time. In your case, you just want the virtual to be a single-mode bundle of e0a and e0b; but you also have an "ifconfig e0a partner e0b" that is probably required.

hemadri37 · ‎03-07-2017

Hi,

Based on what I see above, it looks like vif0 comprises 2 physical interfaces, one of which is down. From earlier posts these are connected to 2 DIFFERENT switches for failover purposes. You currently have e0b down to prevent flapping ?

Both interfaces e0a and e0b are flapping. during this time e0b was down.

It looks like vif0 is in load-balancing mode between the 2 interfaces (multi_mode), which is good for performance, as it load balances across both interfaces (giving you effectively 2 Gig throughput).

vif0: 1 link, transmit 'none', VIF Type 'single_mode' fail 'default'
VIF Status Up Addr_set.

vif is in single mode

would you suggest that vif0: 2 link instead of 1 link as two links are available.

Thanks

Hemadri

hemadri37 · ‎03-14-2017

Dear Members,

The interface connected between Netapp and Switch is a trunk port carrying information of 03 VLANS.

Though Ports connected to both switches from Netapp are flapping at Switch side, storage guys telling ports at netapp are not flapping and no logs seen for flap. out of two interfaces one is up and one is down.

i didnot understand how it happens. also is it require to configure vlans in netapp side also.

can you advise.

pwwiddicombe · ‎03-14-2017

Had you actually read through the document I posted the link for earlier? I believe you need the following to correctly "bind" the 2 interfaces. If set up correctly, the mac address assigned to the pair of interfaces should ONLY send traffic from one link, unless it fails. Now, you also mention 3 vlans - are they all configured to use the same virtual interface, or have they been set up to "optimize" the link by some using one link and some the other (that won't work properly !).

If you want both links to be active simultaneously (as the current Netapp seems to be doing), you need to have a proper etherchannel on the cisco side, with both links into the same switch, or switch stack.

  # second level single vif consisting of both
  # first level vifs; only one active at a time
  vif create single vif10 vif0 vif1

  # use vif0 unless it is unavailable
  vif favor vif0

  # configure the vif with an interface and partner
  ifconfig vif10 `hostname-vif10` partner vif10

hemadri37 · ‎03-14-2017

Hi,

Both links e0a and eob are connected different switches. please ref the pic.

Ports in switch are trunk ports. one interface eob in Netapp is down, whereas in switch port it is flapping and not down. but these changes are not observed from netapp side. they are telling NW issue. i didnot understand what NW issue is there.

csu1a> vif status
default: transmit 'IP Load balancing', VIF Type 'multi_mode', fail 'log'
vif0: 1 link, transmit 'none', VIF Type 'single_mode' fail 'default'
VIF Status Up Addr_set
up:
e0a: state up, since 28Sep2016 16:06:25 (160+16:13:42)
mediatype: auto-1000t-fd-up
flags: enabled favored
input packets 69072059520, input bytes 33990941124602
output packets 94878243386, output bytes 118954255206084
output probe packets 0, input probe packets 0
strike count: 0 of 10
up indications 2, broken indications 1
drops (if) 0, drops (link) 0
indication: up at 28Sep2016 16:06:25
consecutive 13886830, transitions 3
down:

e0b: state down, since 28Sep2016 16:06:25 (160+16:13:42)
mediatype: auto-1000t-fd-up
flags: enabled
input packets 126, input bytes 903580760
output packets 0, output bytes 0
output probe packets 0, input probe packets 0
strike count: 0 of 10
up indications 2, broken indications 1
drops (if) 0, drops (link) 0
indication: up at 28Sep2016 16:06:25
consecutive 13886830, transitions 3

As suggested etherchannel cannot be configured as they are in to two different switches.

pwwiddicombe · ‎03-14-2017

As part of problem determination, I would suggest the following:

1. Disconnect the cable from e0b for a period of time (i.e. 10 minutes). Verify all is normal, and there are no error messages or flapping on either switch.

2. Connect something else to switch B 1/8 (i.e. your laptop). It may function normally (well, possibly, as it's a trunk and the pc would be connected to the native vlan, if you have an operational one), and again there should not be any flapping.

If there is flaps in either of the scenarios above, then the issue is with something on the cisco or cabling side. I'm guessing there are no issues in these scenarios. Further, make sure the Netapp works properly in this scenario; that the vlan tags are properly seen on the Netapp and all 3 VLANS function.

From the last logs, it DOES actually look like server e0b interface thinks it isn't sending any packets on the network. Reconnect it to switch B 1/8 as before, clear counters on 1/8, and observe counters on switch and Netapp side on that link. If Netapp isn't sending any packets, the Cisco side shouldn't see any incoming packets on B's 1/8 at all.

Let us know the results of these tests.

hemadri37 · ‎03-14-2017

Hi,

Sure i will perform this tests and update the results.

Also in this case its not only Switch-B 1/8 flapping, but also Switch-A 1/8 is also falpping.

Switch A - 1/8 connected to e0a of Netapp.

Switch B- 1/8 connected to e0b of Netapp.

both interfaces are flapping continuosly but in Netapp ports eoa is alo not flapping as per the data team.

Regards

Hemadri

hemadri37 · ‎02-15-2017

Dear Mark,

we dont want STP between Switch and server, so we cannot define cost. spanning tree-portfast trunk configured.

issue not resolved. port still flapping. is there anything to do with DTP.

Regards

Hemadri

Mark Malone · ‎02-15-2017

Ok you don't want to use STP then remove one link from one of the switches and connect it to the same switch and put them in a port-channel

as I said you cant physically have a server active/active like that broadcasting to 2 switches that are not bundled in some fashion , its not DTP its the physical design

Think if you have 2 non port channel links standard trunks connected from server to a switch your going to have a loop in traffic and stp will automatically block one , the same thing is happening here only across 2 switches and you need to tell stp to prevent it

one of your links needs to be in standby mode when physically set like that

why would you not want to use STP ?

see this doc on how a loop occurs at layer 2

http://www.omnisecu.com/cisco-certified-network-associate-ccna/what-is-layer-2-switching-loop.php

up to you what you do im just advising you but your config and the physical design you have in place is causing this and stp is the fix or change the design

Switch ports connected to Netapp FAS2020 flapping continuosly