Successful Rockstor 4 install (Virtual Box)

I followed the instructions to create the Rockstor 4 installer - no issues whatever. The instructions are very clear and straight forward. I’m not sure where’s the right place to post this.

When I used the installer in VirtualBox, all went well. I misunderstood the end of the process, where the CLI interface cursor kept blinking, which hasn’t been my experience with other installs. Eventually I logged into the Web interface anyway, and there it was waiting. So I’ve paid my subscription.

The configuration process was the easiest and most straight forward of any of the NAS software packages I’ve tested in VB, and that’s all of them. I forgot to assign serial numbers to the virtual hard disks, so I don’t know if that’s a problem.

I created a Raid 5 pool and a share, and successfully accessed it from my VB host, adding files to the share.

I’m not sure what else I can test in VirtualBox. The final parts for my first, new NAS should arrive tomorrow, and based on my limited experience with a virtual install I’m looking forward to getting it up and working. Congratulations to the development team. I think the result of your work is looking really great.

2 Likes

One thing I did notice,and I’m not sure if it’s an issue: I have about 2.5GB of files in the share I created (16GB total pool usable). I did a scrape and when it finished, there was a “No Csum” count of over 200.

@ceh-u Hello again, and thanks for a very encouraging post. Much appreciated.

To your points:

This is it actually. A fresh forum thread is perfect. Hopefully it will encourage others to post their experience of our first installer for over 2.5 years also. Plus of course the end result is an upstream supported btrfs stack which is nice.

I’m not exactly sure of what you mean here. It may be an artifact of VB. Would be good to know what this looks like so if you could capture a screen grab that would be good. There is a point right at the end of the install, after the initial login prompt has appeared, and as the Rockstor services are finnishin there initial setup, where their [OK] status lines then appear over the login prompt. I tried to account for this by adding the text:

“Enter key for a new login prompt.”

To the end of the first login message. This is pictured in the subsection Wait for the “Rockstor bootstrapping tasks” in our newly added:

Rockstor’s “Built on openSUSE” installer - Beta release

Did it look like that but without the very last “… login:” prompt? As that prompt only appears after pressing the Enter key. As we start the Rockstor service at the end of the boot process it’s service starting reports inevitably end up obscuring the login prompt, hence my ‘work around’ of the enter key prompt. Definitely something we should look to clean up, if possible, but currently I’m unsure of how to do that. Let us know if that is what you mean here.

Thanks for that, again much appreciated and help keep our light on of course.

If it was, i.e. if serial numbers wern’t assigned auto style then the disk page would (read should) complain and you would be blocked from using the affected disks as pool members.

The message that is displayed against each affected device is shown in the following developer focused wiki entry:

Nice; keep us posted. Maybe a fresh thread on your build & install experience. And it looks like you’ve rolled your very own installer just in time. Incidentally, each time you make the installer it incorporates all the latest updates, hence no updates required directly on install. But by the time you do your real hardware install there may very well be some more updates ready so that is another element that would be nice to get feedback on. Especially given this is the only way we can test it, i.e. await real updates to see if they work. The have so far for the latter period of the installers development so they are expected to continue working but there are always going to be hick-ups and stuff we haven’t accounted for so do please keep your reports rolling as it all helps with smoothing our way forward.

For your real install (read for real data), it’s best to avoid the btrfs parity raid levels of 5 & 6. There are a lot younger and less well developed. Stick, at least for the time being, to raid 1 or 10. We will be adding the capability to mix data and metadata levels in the mid term which will allow for stuff like data raid 6 metadata raid1c2 which will help to alleviate some of the known current weaknesses of the parity raid levels. But if you have good backups and use an UPS then the parity raid levels are doable just not recommended for production use. We mention this in the Web-UI tool tips when selecting the raid level for a new pool. Just a note I though best to pop in here really.

No worries, scrubs normally just report this and it’s just an indication of elements of the pool that don’t have associated Csums. Quite normal and not an error condition.

Thanks again for your very encouraging post and for your support. And thanks again for test driving our new rockstor-installer GitHub repository.

1 Like