onboard silicon image raid, problems with

[Normunds]

I'm using onboard Sil3114 chip on K8N Neo-4 Platinum board and I did implement Raid-5 on it. Silicon provides software SATARAID5 to manage the raid(s). There is another onboard NVidia raid controller, I'm using it also, but the prob arrived with Silicon.

What I've done is created 2 Raid-5 arrays across 3 harddrives (the soft allows to allocate just part of a drive and then add them into Raid). It is possible I could have created one big RAID and then partitioned, but in fact my current approach has saved me some ham.

The PROBLEM: at some point I switched off my computer when it seemed to be hanging. That was a fatal mistake as now ONE OF my raids come up with a message. Warning: 'Group 0 "Notes" was not shut down properly' then Error: 'group "Notes" is offline: dirty and reduced'. So one of raids is dead (and the other is running nicely). Well i'm only so glad -- because for the other one -- one that keeps running, i did not have uptodate backups...

So I seem to have lost some data on the first Raid-5 (as my backup was not 100% fresh). Could I still do something? Problems:
1) I see no way to do anything with the failed raid group. It is not accessible -- I would have liked to start it even if some file may not be correct or at least delete it and create it fresh or something.
2) Silicon Image has no forum; they have some kind of "ask question" interface that just swallows questions and sends them to the black hole; their documentation does not talk about this kind of situation (nor it's much of a doc by any standards)

What drives me crazy is -- ALL drives are physically good, still i've lost ALL my data on Raid-5 (or so it seems; OK, minus my backup, but RAID-5 is supposed to have some value to preserve data, eh?). It seems the answer is -- do not use software Raid-5 (unless natively on Linux where you have more say about how it does run). Or it's maybe "do not use Silicon software raid"?

Any ideas how to:
1) start the raid group again (even if loosing some data on it)
2) remove the raid group (just the failing one)
3) get any info/support of Silicon Image controllers/software (in this or other newsgroup/forum?)

Thanks in advance.

[-ChEM-]

its usually safe to just rebuild the raid array with the same config as before

if you have an intact file allocation table, the problem being with the organisation of the RAID, you will have all your files still there

[Normunds]

well, I have no means to rebuild it. SATARAID5 does not offer any option for this array/group. It's not visible in UI, cannot be changed, deleted, inspected, rebuilt -- nothing; just a mention in the log. And occupied disk space marked nice friendly green and not available. And I do not know any means to rebuild the raid group othewise. Could there be other software I could use to manage the raid instead of what is provided by Silicon? One that will understand format used by Silicon (or there is some common standard for raids and there is a good chance that I can find something that will allow me to re-build everything?) I doubt.

All I could do is to take everything (raid) apart -- I guess mark the disks in BIOS as not-raid capable. Launch Windows once like this, then mark them for raid again.
At this point there could be 2 probabilities:
1) Silicon recognizes "raid marks" and re-creates the old setup and we are back at the same place we started with one group functional and other "dirty and reduced", or
2) it sees 3 empty drives. In the latter case I could then recreate my setup (2 raid-5 groups across 3 drives with the same size as before); still at this point it's probably going to be too late.

Or I fail to understand something obvious? What steps would you take to rebuild this array?

[Normunds]

and the solution was:
1) upgrade to the latest sataraid5 version that allowed me to see and delete the "dirty and reduced" group
2) delete it (the tool offered no other options -- I'd have thought it should allow for re-build or something)
3) recreate the group with the same size and type

Really, the files were there, but after I tried to launch a program that used that drive it complained about invalid files or something, so I launched chckdsk that found a lot to do, but still seemed that nothing was obviously lost. So I'm not sure about quality of my recovery, but at least it's not a complete failure.

Finally I left my backup as a current version, added recovered files that I already knew I was missing and saved the salvaged files just in case I find out that I'm missing something else.