@ex3cution Welcome to the Rockstor community and glad you got things going in the end.
Thanks for the report: many good points and very detailed. Some nice pointers re your install finding also.
Thanks, and nice setup.
Smart data is often tricky through a raid controller but it may be possible with the use of custom smart options - explanations and references available via the pen icon next to the smart switch in the disks table. The following in our docs may be of use: “S.M.A.R.T through Hardware RAID Controllers”. Not a polished as I’d like yet but may allow for S.M.A.R.T via hw raid connected disks. All depends on the controller really. There is a canonical link in that doc section to the upstream smartmontools page for this config corner.
I would challenge this as we do nothing custom bar that which is in the main line kernel as per ElRepo’s kernel-ml. It may just be a difference of expectation vs reporting re memory use. Have a look at the following thread where this has come up before and was addressed by forum member @aditaa via a RedHat reference:
Yes, the check summing used by btrfs can be quite a hit. But 2 quad cores should be more than plenty. I know there are improvements going on all the time re efficiency in btrfs and we are inevitably to inherit those as time goes on. Definitely room for improvement and I would no recommend a single core for btrfs. And on the other hand there are times when it fails to distribute across cores the work that is needed. Of note here is that the parity raid levels of btrfs 5/6 are definitely a fair bit worse on this front and essentially impractical currently so as we indicate in the raid level selection tool-tip/hint and our docs - Redundancy profiles section, they are not recommended for production use just yet.
Yes, big one this and we have a rather popular issue open just for it:
Please feel free to contribute to that issue as per it’s intro text by project lead @suman
Very bad and no something we have heard or would expect. I’d look to logs for some clues as this is not normal or reported behaviour so definitely an issue / compatibility / hw conflict of some sort there. You could for instance try a different kernel as it may be down to driver versions or the like. Maybe start a new thread with specific details on this one as we are essentially a re-badged CentOS with a much newer kernel and btrfs-progs.
Yes, translation capability would be great and definitely under consideration, but we have only just started towards that end of things:
https://github.com/rockstor/rockstor-core/issues/821
@Flyer has gotten as far as a proof of concept but I’m not up on exactly how currently (details in that issue, with pics):
And we have the following forum thread with some eager volunteers for the verbal leg work:
again please comment on that thread if you think it’s appropriate.
Bit of a low priority on a headless system but yes; see forum member @henfri’s issue on this:
Yes, we essentially inherit and only really re-badge the Anaconda installer and it is a little buggy. But there are ongoing plans to update our installer to get what improvements we can from upstream, which has only recently had a new 7.4 release tagged:
I know, it’s like magic. But funny thing is a new to raid tec user just expects to be able to do this so doesn’t even notice. But we do have some work to do on other expected behaviour such as changing names on shares etc. But there has been some recent and fairly hefty changes to accommodate such other reasonable expectations in the future.
Had to quote that bit as I feel similarly - but we have a ways to go just yet. However the efforts within Rockstor and the wider btrfs / CentOS / Linux community are perpetually ongoing so I look forward to our future where hopefully all of your concerns are, in time, addressed.
Thanks again for the generous and detailed write-up and hope you get that hard lock sorted as that is not a known issue.
Cheers; and I hope the above references re-assure you that we are ‘under way’ with the various points you raise.