2014-04-29 06:12 AM
After updating from 6.4.2 to 7.0 we see some naming problems with SAN fabrics.
In our fabrics we have the virtual fabric option enabled, but we are using only the default fabric 128. We use additional virtual fabrics for testing new long distance ISLs (10km) or if we have a problem on a existing ISL.
We recreated a new data source to discover our second fabric, because changing the fabric name didn't work. In the data source in the "Advanced configuration" section we set the Fabric Name to: FRD_Office_Fabric2.
When I take a look in the Java Gui at the Zones section, I see the fabric name VSAN 128. Also this name is shown for the fabric in the dwh.
We are using the fabric names to dynamically create service trees for our infrastructure. At the moment I see 5 different fabrics with the name VSAN 128 and not with the name, we set in the data source.
FC Switch Firmware 4.2+ (SSH)
Inventory and Switch Performance are active
We are using the standard options set by NetApp for this data source.
Does anyone have an idea how to fix?
2014-04-30 09:02 AM
I am a little bit surprised this was working for you with 6.4.2, as I believe our Brocade datasource's functionality to "name" fabrics has never played nicely with OCI's virtual fabrics capabilities
Question - something we don't do as it was introduced with FOS 7.0.0 is pull the contents of the "fabricname --show" output from Brocade devices.
If OCI were to automagically name fabrics with the name defined on the switch, would that make your life easier?
Basically, the challenge is that the old idea of letting the user define the name of a fabric for a datasource presumes that the datasource will only ever discover one fabric, which is a flawed assumption in the world of Virtual Fabrics.
2014-05-04 11:38 PM
I think, I know why it works with one of the datasources.
The data source for fabric 1 was created before we have activated the virtual fabric option on the switch. Because of this he safed the name of it.
To get the information of the fabricname --show would be a good alternative for us.
How long would it take to get a brocade data source with this command enabled?
2014-05-05 04:29 AM
As this seems to be a relatively small change, as well as being very low risk, I opened this as a minor improvement bug with engineering - this means engineering may take a shot at this during our "shift" rotation, where some of our engineers work on bug fixes / minor improvements for a month, instead of larger projects (such as entirely new datasources, infrastructure work, etc). However, this makes it difficult to forecast when we may be able to do this, as improvements are below bugs in priority.
I will set myself a reminder to look into this for you by the end of the week
2014-07-16 07:57 AM
I know it has been a big week for Germany, but it just got a little bit bigger.
Do you have a lab OCI 7.0.0 system you would like to test a patch on? We cannot simulate this on existing OCI datasource recordings as we are going to issue a new fabricname FOS command in the context of each VF we discover.
Right now with our test patch, we are using the fabric name exactly when it exists - we are thinking of including the "VF: NN" as well, to maintain visibility of the VF number. In the screenshot below, I have only named 3 of the 6 VF on this DCX
Edit: if you click the image, it should appear larger, and not blurry.
2014-07-16 10:42 PM
Thank you :-). Yes, was a great week.
We can test it in our normal environment. Do you know, is the name also transferred into the DWH?
I will send you my email address via pm.
2014-07-17 03:54 AM
Yes, this will be reflected in the DWH. I am taking a quick look at the dwh_inventory mart, and I'd expect this to change what is seen in the "name" column in the fabric table in the dwh_inventory mart