I have software developed in CLIPPER, on the TRUENAS CORE server my Windows 2003 Server 32bits accesses the application and runs normally, I created another server, this one now using TRUENAS SCALE DragonFish with the same access rights as TRUENAS CORE, but the application does not work , it doesn’t run, no message appears, both machines have the same motherboard and memory configuration, the only difference is that on the TRUENAS CORE server I have 2 1 terabyte HDs doing RAID 0, on the TRUENAS SCALE server I have 3 500 Gigabyte hard drives doing RAID 5, I’ve already reviewed the entire configuration of the two servers and I couldn’t find anything that could cause this type of problem, does anyone have any idea what could be happening.
I am not sure that what you are saying is the ZFS style of RAID or something your motherboard or add-in card is configured to do.
As others will surely mention soon after my post, you must give direct access to your drives to ZFS without any kind of RAID setting, and best provided by an HBA card. This direct access I believe is a passthrough mode where the controller or motherboard does ZERO cache or management of the drives. It may also be ‘raid 0’ mode although I am uncertain.
The more seasoned NAS folks will offer high quality comments/support and probably point you to various documentation to support the correct way to handle drives in your box(es).
I agree - you need to provide more information about your TrueNAS SCALE build than you have so far.
Also, I am assuming that you mean that this is a CA Clipper (v5?) database application.
I do not have any experience with Clipper, and you should have no difficulties with storing the Clipper executables / code on your TrueNAS system, HOWEVER you should be VERY wary about storing the Clipper database on your TrueNAS server.
TrueNAS (both CORE and SCALE) use ZFS as the file system, and ZFS is a “Copy-on-write” file system, which means that (unless you use a zVol to hold the data file) literally EVERY time you update a few bytes in the file, ZFS takes a full copy of the file in order to make the changes. In other words, it is entirely unsuitable for hosting file-based database files such as MS Access, Clipper etc.
No, this isn’t the way ZFS works–it’s a fresh copy of the block, not a whole new copy of the file. iSCSI/zvol may still be the better way to go (or it may not; I’m not in the least familiar with Clipper), but not for that reason.
3 Likes
Ah - yes I can see that now.
So I guess that the block size choice is important.
Think so. Remembering that what really matters is the block size after compression. And the spans across disk.
You probably want mirrors.