- May 29, 2011
So what a lot of people forget about all of this, because they're used to things working "well enough," which is fine in the Windows world where a BSOD is fine now and then, is that it gets really difficult to get those last 9's in the '99.999_% reliable'.Can anyone elaborate on this or point me to more info? What kind of issues? Performance? Trashed data? It sounds like setting up FreeNAS in a VM and verifying it can see and interact with the drives on the passed-through controller isn't sufficient to prove pass-through is working correctly, so are there specific tests that can be run to determine if my particular system is prone to these problems? Or is it always going to boil down to "run it for months and see if it blows up"?
I recommend hardware based on observations and experiences that suggest a certain bit of gear is more likely to work. I pick operating systems based on reliability and code quality. Here in the forums, as an example, there are lots of ways you *can* attach your drives to your host, but if you look carefully, you'll find that the favored one is an LSI HBA with a certain revision of firmware, combined with a certain driver. This isn't because other combinations might not work, but rather because there are probably BILLIONS of hours of positive experience with that combination of HBA/firmware/driver. This is not a guarantee that if you get that same combination that you'll experience trouble-free operation, but it does stack the deck in your favor. Given two imperfect things, I try to prefer the better of the two, the one that's been less problematic, etc.
The only way to *really* gain confidence that things work is to test the heck out of them. When we build gear here in the shop, it runs for at least a month in burn-in, looking for problems or faults.
When you try some unusual configuration of FreeNAS, you lose out on the benefit of millions of run-hours of all the users who have an X9SCM-IIF and an IBM ServeRAID M1015 running in a chassis for the last half dozen years, a configuration known to be as close to problem-free as can be.
Doing something else isn't a guarantee that you'll have a problem, or that it won't work, but I can tell you from experience that you do lose out on the experience benefit offered by that "millions" and you suddenly find yourself in the "dozens" (a few days of testing) or "hundreds" (a month of testing) or maybe even "thousands" (three months of testing) range of test-hours, because it is just your rig and it is up to you to find the sharp edges. I start to get more confident out past the "thousands" mark.
Still, as a cautionary reminder, consider that I kept a hypervisor on ESXi 5.5 and FreeNAS 8.2 for many years in order to avoid the inherent risks of frequent updates, and only just recently updated it to ESXi 6.0 and FreeNAS 11.1-U4 after some extensive testing. After moving the disks, I *still* ran into a platform issue that resulted in a disk dropping and needing to resilver. Any time you're virtualizing you're increasing the complexity of what is going on fairly substantially, and fate is there hiding under the covers waiting for you to make a fatal mistake.