Revodrive X2

[no_connection]

The benefit from booting FreeNAS on my "workstation" is that it detected my Revodrive X2.
It detected all four 60GB drives so I believe it might work just fine. I can see SMART for he individual drives in windows so the controller appears to pass that through.

Knowing there have been quite some problems with the Revodrive, bad sectors and other misbehaving (at least with older firmware, so it might be fixed now) what is the best way to use it?
I guess it could be the RAID controller making all the problems as I didn't find any when using passthrough/HBA mode.
Although windows detected it about half the time but that could be due to the MB being flakey.
And considering I will not be replacing it if it breaks and there is no way to individually replace the 60GB disks what would be the "optimum" way to configure it.

What would be the worst thing that could happen if it turns out to disconnect on boot or start generating errors?

Using it as .system dataset?
ZIL?
Jails?
ESXi store?
Cache when editing movies?

 

[no_connection]

I get why l2arc is not a good idea. but ZIL(SLOG)?

I was hoping FreeNAS itself could replicate(snapshot?) it to a datastore in my main pool, or at least any important data. It would not be that much data.

If I would use it in my workstation it would be striped and used for cache or files needing good IO or IOPS, not for irreplaceable data.

It might not be redundancy from a complete failure, but if I mirror or Z1 I should at least get data integrity and know if something went wrong.

So I'm only worried about it dragging anything else with it, which might have different consequences depending on how I use it, which is why I posted.

*edit* now it just looks like I'm replying to my self...

 

[anodos]

If you use it as a zil (which you probably don't need), then it will most likely take your pool with it when it fails.

Also I'm not sure how much real data integrity you would get by mirroring them. Since they are SSDs experiencing same write load and manufactured by same vendor at same time. They will die pretty closely together.

A better question to ask is whether you need the IOPs for anything server-side since your network is probably your bottleneck anyway. It sounds like you have a solution in search of a problem.

 

[Ericloewe]

If you use it as a zil (which you probably don't need), then it will most likely take your pool with it when it fails.

Also I'm not sure how much real data integrity you would get by mirroring them. Since they are SSDs experiencing same write load and manufactured by same vendor at same time. They will die pretty closely together.

A better question to ask is whether you need the IOPs for anything server-side since your network is probably your bottleneck anyway. It sounds like you have a solution in search of a problem.

Pools should survive bad SLOGs (obviously with loss of everything contained in them that hadn't been properly written yet).

 

[no_connection]

Also I'm not sure how much real data integrity you would get by mirroring them. Since they are SSDs experiencing same write load and manufactured by same vendor at same time. They will die pretty closely together.

A better question to ask is whether you need the IOPs for anything server-side since your network is probably your bottleneck anyway. It sounds like you have a solution in search of a problem.

Data integrity, as in knowing the data is not corrupt would still be there, and with the ability to correct should one drive give a bad block.
I know that data availability (loosing the entire thing) will not be mitigated, but that is what backups are for.
I would say that I have a disk in search for use as it does not work properly with the MB in my workstation.

IOPS/latency over network would still crush spinning rust in performance. (I should test how a simple windows file share preforms, it can't be worse than that right?)
So if nothing else I could use it to load portable application over network instead of local disk. (I love portable apps)

If safety of my main pool can not be compromised I might just test it.

 

[Ericloewe]

Data integrity, as in knowing the data is not corrupt would still be there, and with the ability to correct should one drive give a bad block.
I know that data availability (loosing the entire thing) will not be mitigated, but that is what backups are for.
I would say that I have a disk in search for use as it does not work properly with the MB in my workstation.

IOPS/latency over network would still crush spinning rust in performance. (I should test how a simple windows file share preforms, it can't be worse than that right?)
So if nothing else I could use it to load portable application over network instead of local disk. (I love portable apps)

If safety of my main pool can not be compromised I might just test it.

Why are you bringing up backups when we're talking about an SLOG? You do realize data is in there for a couple of seconds at most (except if something goes wrong and it's actually needed)? You'd lose the few seconds' worth of data that was being transferred. Unless you're backing up the data somewhere else before copying it to FreeNAS, your backups are useless in retrieving that bit of data that was lost.

 

[no_connection]

It has four 60GB drives attached to a PCIe controller. So the other two (if I where to mirror two for SLOG) would still be free for other uses.
But SLOG would only be used if it is safe.