08-17-2011 03:49 AM
I need to replace a brocade 5000 fiber channel switch (in existing fabric) with a newer model with more ports.
How should the switches be connected to each other to replicate the zonig database?
Does anyone maybe have a "how-to" for exchanging fc switches?
Thank you in advance
08-17-2011 08:02 AM
it is difficult to give any advice with such a little information. If you only worry about the zoning information then you can establish an ISL between both switches and then you have it on both boxes if the fabric will merge if not you have a segmentation.
Both events are visible to the servers and cause IO stop.
I hope this helps,
08-17-2011 08:57 AM
The so called FabricBuild Event forces all traffic to be stoped. A couple of infromation has been exchanged through the entire fabric like NS and routing table updates. As well it has to be checked if there are any conflicts between the switches / fabrics.
During this time no traffic will be processed.
After the fabricbuild has finished sucessful the frames will be transported.
This information was given during a Brocade FC training. If this is currently no longer true please let me know.
11-18-2011 12:39 AM
I didn't get any notification to my topic, that's why I lost touch with it.
I try to provide a bit more additional information, now that I have read a few more postings.
We only use soft zoning and do not have any trunking enabled all hosts are connected to both fabrics for full redundancy.
The new Brocade 5100 (replacing the Brocade 5000 switches) are yet configured with the full fabric and PoD licenses as well as a new, unique domain ID, IP configuration, NTP server and a switch name.
The switches are already rack mounted, powered on but disabled and not yet connected to the fabric.
What I'm now planning to is:
clear the new SWs configuration (cfgclear)
clear statistics and log information (portstatsclear, portlogclear)
establish an ISL then reboot the switch (noning DB should be transfered automatically as well as joining the switch into the existing fabric)
Further steps would be:
Make a backup of the new switch configuration
unplug the FC cables from the old switch and plug them into the new switch
disable the old switch, remove the ISL physically remove it
Hopefully, that's it then.
The fabric.ops parametes are all default - in my opinion but how can I be sure?
Did I forget anything?
Thankful for any useful input..
11-20-2011 03:08 AM
so far looks OK.
You have nothing written about your server and arrays which are affected. Are all hosts down?
In case of all hosts are dwon I would do it in different way.
On the new SAN switch disable all ports with portcfgpersistentdisable first.
Switch off the old SAN switch (switchdisable)
Configure the new switch with the DID of the old switch and connect it to the fabric. (cfg is removed on the new switch and allaccess is enabled)
Enable the new switch and run the portcfgpersistentenable command on the ISL port and check ISL. Check with fabricshow that the new switch has joined the fabric.
Check with the errump command for any errors.
Second I would plug the hosts and storage arrays on the same ports as it was on the old switch.
This would avoid any issues with dead devices on the OS side if AIX or persistent bindings are in place.
Enable with portcfgpersistentenable the array ports. Check that no errors are seen on the ports. After enable clear the stats and check for any new errros some minutes later.
Enable the server ports and power them on. DO the same checks as on the storage ports.
Crossing fingers and wait until you get feedback from the OS admins.
If you are happy with this please rate the thread,