2017-10-27 09:03 AM
I have a situation to do an image upgrade on a couple of nodes that's not in our network anymore. The only option is to connect to the console port, setup the cluster and upgrade the image. I already completed a boot option-4 (clean up) on both nodes.
The challenge now is how to transfer the image from my laptop to the controllers using the http server on my laptop meaning that my laptop NIC should be able to talk to a port on the controllers (e0a, e0b, e0M etc). I can setup any of these ports on the controllers with an IP address (same subnet as my latop IP) during the initial cluster setup but I am wondering how to transfer the image to the controller using the http server when my only one laptop NIC is connected to the console port on the controller. Will I be able to get this job done using a small 4-port switch.
I am not sure if my laptop NIC will be able to talk to the controller console port through this switch. It will be a great help if I could get an answer how to make all these cable/port connections on to the 4-port swtich and to get the image upgrade done before I ship it out to another location.
2017-10-30 07:44 AM
So basically the connections are as below as I understand it.
Console port to a port on the switch
e0M to any port on the switch
Laptop NIC to any port on the switch
and use regular ethernet patch cord for the cable connections. Not sure the console port will work this way though. I tried but I was not able to get in to the console session. Can anybody confirm if it is supposed to work this way?
2017-10-30 09:07 PM
2017-10-31 09:28 AM
Sorry, that's my mistake. Actually the console port goes to the USB port on the laptop using an USB adapter and I should be able to connect the network port directly to e0M without the need for a switch. I totally forgot about it. Thanks for reminding that.
2017-11-02 08:17 AM
The console port + e0M NIC connection didn't work for me. The laptop and the controllers were not able to talk to each other due to some reasons. The way it worked for me is by using option - 7 from the boot menu and transfer and upload the image. However, I don't think this is a good way to downgrade to an older version of image for production systems because I had issues during the reboot. The system failed to detect a root volume and went in panic mode. It was okay in my case to wipe all the drives and rebuild the system with the new image. I got the job done anyway but I would like to encourage further discussions on what went wrong during my attempts to talk to the controller's cluster/node shell through e0M or the other ports in order to copy the image from my laptop. I tried various methods using direct cable connections, using a switch etc. without any luck.
2017-11-02 08:34 AM
Using option 7 for reverts may work for going down a P release, but when you cross a version boundary you need to first issue the revert commands from within ONTAP. Otherwise, the disks will have labels the old code will not understand, features may be enabled on the cluster the old code can't support, etc.
2017-11-02 08:53 AM
I used to do this a lot in the field. I kept a USB powered switch in my toolkit just for this purpose. I would connect my laptop and all the e0M ports to my staging switch, then go node by node with a console cable running SP Setup at the LOADER prompt. Then I would connect to the SP ports over the network (which shared the e0M port on those platforms), use the option 7 menus to load any required code updates, and run through cluster setup.
2017-11-02 11:36 PM
The way it worked for me is by using option - 7 from the boot menu and transfer and upload the image. However, I don't think this is a good way to downgrade to an older version of image for production systems
You never said anything about downgrade. You said that you were going to reinitialize filer and then upgrade to new version.