My OS drive (USB thumb drive) was reporting errors so I’m working on getting a new installation up and running. I have a few questions regarding this procedure…
First, after I enable stable updates, is there a good way I can trigger the update via SSH so that I may monitor the process? After my install last night, I started the update process from the GUI. After about 90 minutes, it was still updating. I wasn’t sure if it was hung or not, so I SSH’ed in and issued a “sudo reboot”. My thought was that whatever was happening, it should shut down gracefully. What actually happened is that ‘yum’ was left in a broken state. I “fixed” it manually with yum-utils, but I’m thinking that I should just reinstall again to avoid any potential issues. From looking on the forums, I’m leaning towards the appropriate command (running it via ‘screen’ in case I get disconnected) being something along the lines of
yum update && yum upgrade && yum update rockstor
Second, I think I also read on the forums that when a config is loaded the appropriate rock-ons will be installed. Is this the preferred method of reinstalling the rock-ons? I think I had an old Plex installation active as I think there was a bug that kept it from being easily updated. I’m assuming that the latest docker will be used to reinstall things and not and old one. Is that an accurate assumption?
Lastly, one of my installations failed because the OS drive’s file system was LVM. Is there a way to modify the installer so that LVM cannot be used at all? I briefly considered doing a manual partitioning of the drive, but backed out and did an automatic partitioning instead. However, LVM was used and the GUI through an error when I tried to set the hostname and create the initial user. I haven’t verified that these steps will reproduce the issue, so I haven’t opened a ticket for it. I don’t want to make anyone go off chasing a ghost.
This has been seen a number of times. Basically leave the rpm db a little scrambled and can be tricky to repair. Partly down to our really old ISO but we should be fixing that soon with our “Built on openSUSE” based installer.
should be enough really.
This is only if the config backup was taken with a fairly recent version of Rockstor; this feature was added by @Flox in 3.9.2-52:
and the restore would, likewise, have to be performed post upgrading to at least this version.
Yes, should be.
Yes, only the ISO’s kickstart configured default btrfs arrangement is supported.
Our as yet unreleased installer tare of this. It’s image based so will just establish a supported partition / filesystem arrangement on the chosen drive and that’s it. Much quicker and much easier. Our prior installer was based on a re-badged CentOS anaconda installer. We have now departed from this arrangement entirely and will be using a kiwi constructed installer setup to configure the system with no ‘user adjustable’ options as we are attempting to achieve an appliance type approach where all config is via the rockstor Web-UI.
Hope that helps and answers your questions. And thanks for the report. We really need to get to releasing this new ISO but it has to coincide with our new Stable channel release on our Built on openSUSE offering and that’s just not quite there yet.
Thanks, at this point the only remaining issue is that the Rockstor images aren’t showing up as installed in the rock-ons page. I’m removing them with "docker container rm " and then installing via the GUI. Plex did install as an old version, not the latest, but it is quite possible that I intentionally did not specify “latest” when I did the original install a year or so back.