Start a Conversation

Unsolved

This post is more than 5 years old

1 Rookie

 • 

20.4K Posts

1798

August 29th, 2010 20:00

How to force FSN failover

Hello,

is there a way to force FSN failover from primary to standby ? We had a situation last week where cge0 (primary) was connected to switch A and cge3 was connected to switch B, switch A lost its communication to the core switch. So there was never a link down event and FSN never failed over. Can i remotely force FSN failover ? I've looked at server_sysconfig but nothing jumped out at me.

Thanks

August 30th, 2010 02:00

You could set the port down on the network switch, which would trigger a FSN failover to your other switch.  With regards to a command on the Celerra, I don't see a command I can provide to do it.

8.6K Posts

August 30th, 2010 03:00

There is no customer facing command to do that directly.

You can try to ifconfig down the physical interface and see if that does the trick

support should also be able to help you with some debug commands

Please open a product enhancement request (PER) via Powerlink if you have a minute

Rainer

1 Rookie

 • 

20.4K Posts

August 30th, 2010 04:00

ifconfig does not work on physical interfaces, support and I have tried that when we had another issue where DART stopped working on a physical interfaces, while network switch was still showing good connectivity. I'll open a PER (not holding my breath, it's been 4 years now that i filed a PER for Celerra commands exit status to be published, ridiculous !)

1 Rookie

 • 

20.4K Posts

August 30th, 2010 04:00

Kevin,

i don't have access to network switching infrastructure so have to do it from Celerra side.

Thanks

366 Posts

August 30th, 2010 05:00

Hi,

there are maintenance commands to shutdown an specific device ( cgex, fgex, fgxy, etc...), but unfortunately they are not customer facing.

If you open a service request, the tech support, or your local field support can assist you.

Gustavo Barreto.

8.6K Posts

August 30th, 2010 06:00

Hi dynamox,

sorry – I keep forgetting that you can only shutdown an interface but not a device with server_ifconfig.

There are maintenance commands to look at and manipulate a FSN which support should be able to help you with.

Rainer

99 Posts

August 30th, 2010 07:00

It MAY be possible using the server_sysconfig command if you do the following:

1 - Hardcode your switch to the desired speed (ex: 1000FD)

2 - Use the server_syconfig command to hardcode the primary ports of the FSN to a different speed other than auto/auto: server_syconfig server_2 -p cge0 -o "speed=100,duplex=full"

This should cause your primay links to go down,  and therefore cause the others to failover.

Let me know if this helps.

- Sebby Robles

August 30th, 2010 09:00

Thanks for the tip.  I'm hoping for a solution that's "hands off" on the switch side, and I think this is a good start.  I have no physical access or contact with the folks responsible for the switches connected to an NS42G that I'm supporting - it's another department's Celerra (their admin is out on maternity leave).  Nobody in that group can even tell me who to call about the switch gear.  Even since the Force10 switches have had this (buggy?) code loaded, I get flapping errors.  I wanted to failover to the old Cisco switch on the standby side of the FSN.  This is a good idea, so thanks much for suggesting it!  I will try it the next time I see more errors from the Force10.

Thanks much!

Karl

1 Rookie

 • 

20.4K Posts

August 30th, 2010 09:00

i don't know about Celerra but on a server side (HP Blades with broadcom cards) if i have duplex/port mismatch, the link never goes down it simply stops working so i am not sure that would trigger FSN failover. I will be installing new ProCurve and NS120 in a few weeks so will test it out. In my main datacenter i don't have access to switch gear so i will not always be able to verify port settings on the switch side.

99 Posts

August 30th, 2010 09:00

Karl

It is best practice to have the switch and Datamovers hard coded to the same speed, so chances are the switch is already set to 1000FD.  If you have a flapping  issue, this is most likely due to a speed/duplex mismatch.

Set your ports to auto/auto, and then verify what speed they are connecting at.

Let me know if this helps.

- Sebby Robles

August 30th, 2010 10:00

Ah, if only it were so simple!  In this case, the flapping is caused by incompatibility between the Force10 and an upstream Cisco switch.  The Cisco hosts than the VLAN for this unit's NS42G.  I don't have all the details, but the unit admin told me they had problems with the Force10 reporting it saw the MAC address of cge0 in the wrong place.  They upgraded the code on the Force10 to fix it and the interface only flaps under load.  The duplex settings are all forced 1000FD, so they don't think that's the problem.  This problem seems related to the code and being worked on separately.

Thanks!

8.6K Posts

August 30th, 2010 12:00

please make sure to file PERs for more control over devices and FSNs

Rainer

No Events found!

Top