Complete beginner to FC... need a little help

[Anonymous]

Hey,

We've purchased a few Brocade 300 switches for the storage network for our new virtual infrastructure. We've got them setup as follows, so each host has two HBA's connected to two different FC switches. These are then connected to another FC switch in 'storage rack' via two cables. These storage FC switches then have two links to each of the NetApp controllers. As per the diagram below.

So are as far as I can figure out, we basically need two fabrics, cluster-FC1 and storage-FC1 in one fabric, the other two in another? I've set different switchID's on each switch already but a switchshow throws up some zone conflict errors. Ports 0 & 1 are the two ports that link to the other FC switch.

Index Port Address Media Speed State     Proto

==============================================

  0   0   650000   id    N8   Online      FC  E-Port  segmented,(zone conflict)

  1   1   650100   id    N8   Online      FC  E-Port  segmented,(zone conflict)

  2   2   650200   id    N8   No_Light    FC

  3   3   650300   id    N8   No_Light    FC

  4   4   650400   --    N8   No_Module   FC

  5   5   650500   --    N8   No_Module   FC

  6   6   650600   --    N8   No_Module   FC

  7   7   650700   --    N8   No_Module   FC

  8   8   650800   --    N8   No_Module   FC

  9   9   650900   --    N8   No_Module   FC

10  10   650a00   id    N8   Online      FC  F-Port  50:0a:09:84:8d:2c:b6:ed

11  11   650b00   id    N8   Online      FC  F-Port  50:0a:09:83:9d:2c:b6:ed

12  12   650c00   --    N8   No_Module   FC

13  13   650d00   --    N8   No_Module   FC

14  14   650e00   id    N8   Online      FC  F-Port  50:0a:09:86:8d:2c:b6:ed

15  15   650f00   id    N8   Online      FC  F-Port  50:0a:09:85:9d:2c:b6:ed

Am I right in thinking that we need a ISL trunk setup between the two FC switches?

Basically I'm coming to this completely fresh having worked with Cisco gear in the past and so I'm just wondering if anyone can outline an approach of what would be needed here?

I've set different SwitchDomains for each switch, but do you need the switches in the same fabric to use the same SwitchDomain?

Do you need any other configuration on the ports that the NetApp or the Hosts are plugged into?

Any tips/hints/pointers or links to begineers guides would be greatly appreciated

Cheers,

Andy

[Anonymous]

Howdy!

Am I right in thinking that we need a ISL trunk setup between the two FC switches?

>> Not necessarily, but it is beneficial to have a trunk.

Basically I'm coming to this completely fresh having worked with Cisco gear in the past and so I'm just wondering if anyone can outline an approach of what would be needed here?

I've set different SwitchDomains for each switch, but do you need the switches in the same fabric to use the same SwitchDomain?

>> No, one suggestion is to have unique ids even between the fabrics. So that the domain ids in one fabric is 1,3 and in the other fabric the domain ids are 2,4

Do you need any other configuration on the ports that the NetApp or the Hosts are plugged into?

>> The ports are logged in (based on the switchshow)

Any tips/hints/pointers or links to begineers guides would be greatly appreciated

>> The switchshow states quite clearly what the issue is: the zoning databases are not identical between the two switches. You can either manually make the zones identical, or more easily just clear them on one and when you connec them the zoning db should be distributed also to the new switch. You also need to make sure that the default access mode is similar to the switches: see the CLI command 'defzone'.

To clear the config you could do cfgdisable;cfgclear;cfgsave or configdefault -all or both.

[Anonymous]

Cheers for the reply, I'll take a look at setting up a trunk between the two switches then but again if you know the commands off the top then that'd be appreciated

I've actually set the SwitchDomains to totally different values anyway so should be ok there.

>>>> The switchshow states quite clearly what the issue is: the zoning databases are not identical between the two switches. You can either manually make the zones identical, or more easily just clear them on one and when you connec them the zoning db should be distributed also to the new switch. You also need to make sure that the default access mode is similar to the switches: see the CLI command 'defzone'.

>>To clear the config you could do cfgdisable;cfgclear;cfgsave or configdefault -all or both.

I've done that on the two connected switches but they still show zone conflict. Will the trunk need to be sorted first?

Also the default access mode is the same on both switches.

Default Zone Access Mode

        committed - All Access

        transaction - No Transaction

[Anonymous]

Disabling and re-enabling the port may have actually sorted my problem

14  14   c90e00   id    N8   Online      FC  E-Port  10:00:00:05:33:a2:31:d6 "storage1-FC1" (upstream)

15  15   c90f00   id    N8   Online      FC  E-Port  10:00:00:05:33:a2:31:d6 "storage1-FC1"

So I can see the switch that's attached.

Just need to config the trunk up now.

Do trunks have different options? So will it load balance over the two links, or prioritise the one and then the other?

[Anonymous]

Trunk don't really have an option and form automaticly if

-the ISL's have the same portsettings, with regards to speed QoS etc.

-ISL's are on the same portgroup

-Difference in length between ISL's is no more then 300 meters, but a difference over 30 meters could introduce an perfomace issue.

Load balancing is done by means of exchange based routing (EBR) or port based routing, depending on your switch setting, the default is EBR.

In the old days ISL's in a trunk had equal amounts of traffic going through each ISL, nowadays you'll see a difference per ISL.

That doesn't really matter as the traffic will overflow to the other ISL's if its needed.

Btw i advise you to fix the portspeed for every port. Reason being that optics can degrade over time, should that happen AN will trotlle back the speed on a level were it functions again. Until the optics have degraded even more and AN will throttle back again. Is this behaviour would go unnoticed you may end up with weird performance issues. By fixing the speed it will either work or it doesn't. When the latter happens you know you have to replace some parts.

If you have an monitoring system to check for error counters to be able to do a preventive swap than that even better.

[Anonymous]

You could also check 'portcfgshow'. Is trunking enabled on all ports?

I'm unsure, do you need a trunking license on these switches?

I suppose it should show up in 'licenseshow'.

you check which routing policy you have with 'aptpolicy'.

[Anonymous]

Hi,

I am also facing same problem. However disabling and enabling port has not helped me to resolve the issue.

can anyone  help me to resolve this issue?

Thank you.

Chandan K P

[doges]

a trunk only forms if some prerequisite are met

-trunking license

-trunking on the ports taht should form a trunk is enabled, use portcfgshow to verify

-isl ports that should form a trunk need to be on the same portgroup of the trunk example port 0-7 or 8-15 or ...

-same portspeed

consider before enabling isl/trunks if you want to enable encryption and/or compression