2016-02-02 08:21 AM
If there is a good thing that can come up from any incident, is showing us our SPOF (Single Point of Failure).
Long story short, I have an FAS3240 in co-location and yesterday it crashed. Don't worry, everything is back to normal now. But, during the incident, I wasn't able to access the SP remotely, and since the filer was in takeover, I couldn't check the configuration of the SP.
So, I'm wondering if some of you have network device to get access to the wrench port remotely? Just like in the days with the HP Jetdirect print server device.
I've looked at the Digi products (http://www.digi.com) but I'm not sure which one I should get.
Thanks for you help guys!
2016-02-02 11:44 AM
Hi René -
I'm not sure what you're trying to accomplish here.
The "wrench" port on a FAS 3240 is a network port which presents both e0M and the SP connection on the same port. There is an internal switch behind the wrench port, so the implication is that both e0M and SP have to be in the same broadcast domain. If the SP is properly configured with an IP address/mask/default gateway and the wrench port is connected to the proper network segment/broadcast domain, it just works whether the node is in failover or not.
There is no need to have an external network device to connect to the SP through the wrench port since it is already a network interface. The console port is also available which is a pure serial connection, but that's different than the SP also.
Hope this helps.
Lead Storage Engineer
Kudos and accepted solutions are always appreciated.
2016-02-02 09:01 PM
2016-02-02 09:44 PM
I've used assorted serial port concentrators (digi included) to get remote access to devices in infrastructure I've deployed in the past, but not generally to something like a Service Processor that was already providing out of band management. The exception would be where there was a design requirement to have modem level access to an environment in the event of a total loss of network connectivity.
Unless you fall into that relatively extreme scenario, I would design redundancy into the switch paths so that your in-band access (e0a, etc) and your out-of-band access (SP/e0M) are via diffferent switches or network paths. Also look at things like connecting node-a SP and node-b SP to different switches. This protects you from assorted switch failure scenraios.
I've also on occasion run the serial consoles to nearby servers just in case. Often a nearby management server will have console ports available. Managed PDUs with port level power control can also get you out of some fault scenarios when the installation is just too remote, too innaccessible, or simply unmanned.