danb35
Hall of Famer
- Joined
- Aug 16, 2011
- Messages
- 15,504
I decided to give SCALE a try today. Big mistake. Fortunately, I installed it to its own SSD and was able to revert.
I'm currently running CORE 12.0-U8 (not U8.1), and I was trying to install SCALE 22.02.0.1--I'd apparently missed the download for 22.02.1. My plan was to download the config file from CORE (including the password seed), install SCALE to its own SSD, boot, upload the saved config, and (so I expected) be on my way. Not so fast.
My first problem came with configuring the network interface. Under CORE, I had the two ports of my Chelsio T420 in a LAG group as lagg0, with a static IP assigned. I expected I'd boot with those ports unplugged, manually set up the LAG group at the console menu, assign its IP address, and then be able to access the GUI. No dice there. The first problem is that it needs to be called bondX rather than laggX. Of course, nothing in the console menu tells you that until you try to save it with the incorrect name, but that's easy enough to fix. The second problem is that, AFAICT, there's simply no way in the SCALE console menu to assign a static IP address to an interface. It was there in CORE, but not in SCALE. All you can do in SCALE is turn on DHCP for one (and only one) interface (and you have to manually do that, as DHCP isn't enabled by default on any interface). I ended up doing this, then setting the static IP in the GUI--awkward, but it worked.
So I'm now able to log in to the web GUI, and true to their word, iX have built a completely different interface without bothering to backport it to CORE. And they want us to believe that CORE will remain a viable product into the future... Anyway, once logged in there, I'm able to find the place to upload the saved config file, I do so, and the system reboots.
...and here's the problem. Once it reboots, toward the end of the boot sequence, it gives a bunch of errors which disappear far too quickly to make note of them, then clears the screen and gives a message that the middleware isn't running, and to press Enter for a root shell. I press Enter, and that's exactly what I get: a root shell. The network is down, I see the bond0 interface doesn't exist, and as advertised the TrueNAS middleware isn't running, so there's nothing I can do without digging around with more Linux internals than I really care to on this box. Since I didn't want to continue to have the NAS down with no idea of what to do on it, I powered it down, de-racked it (again), disassembled it to access the SSD tray (again), and reinstalled the old SSD with CORE on it. It boots, and other than suddenly being unable to communicate with the UPS, it's fine.
So what's going on here? Should this have worked? Is the LAG interface likely to have caused the problem? What can I do to track it down--ideally without tearing down the server several times to swap SSDs?
I'm currently running CORE 12.0-U8 (not U8.1), and I was trying to install SCALE 22.02.0.1--I'd apparently missed the download for 22.02.1. My plan was to download the config file from CORE (including the password seed), install SCALE to its own SSD, boot, upload the saved config, and (so I expected) be on my way. Not so fast.
My first problem came with configuring the network interface. Under CORE, I had the two ports of my Chelsio T420 in a LAG group as lagg0, with a static IP assigned. I expected I'd boot with those ports unplugged, manually set up the LAG group at the console menu, assign its IP address, and then be able to access the GUI. No dice there. The first problem is that it needs to be called bondX rather than laggX. Of course, nothing in the console menu tells you that until you try to save it with the incorrect name, but that's easy enough to fix. The second problem is that, AFAICT, there's simply no way in the SCALE console menu to assign a static IP address to an interface. It was there in CORE, but not in SCALE. All you can do in SCALE is turn on DHCP for one (and only one) interface (and you have to manually do that, as DHCP isn't enabled by default on any interface). I ended up doing this, then setting the static IP in the GUI--awkward, but it worked.
So I'm now able to log in to the web GUI, and true to their word, iX have built a completely different interface without bothering to backport it to CORE. And they want us to believe that CORE will remain a viable product into the future... Anyway, once logged in there, I'm able to find the place to upload the saved config file, I do so, and the system reboots.
...and here's the problem. Once it reboots, toward the end of the boot sequence, it gives a bunch of errors which disappear far too quickly to make note of them, then clears the screen and gives a message that the middleware isn't running, and to press Enter for a root shell. I press Enter, and that's exactly what I get: a root shell. The network is down, I see the bond0 interface doesn't exist, and as advertised the TrueNAS middleware isn't running, so there's nothing I can do without digging around with more Linux internals than I really care to on this box. Since I didn't want to continue to have the NAS down with no idea of what to do on it, I powered it down, de-racked it (again), disassembled it to access the SSD tray (again), and reinstalled the old SSD with CORE on it. It boots, and other than suddenly being unable to communicate with the UPS, it's fine.
So what's going on here? Should this have worked? Is the LAG interface likely to have caused the problem? What can I do to track it down--ideally without tearing down the server several times to swap SSDs?