All-Flash pool checksum errors?

[Doug McIntyre]

Hi there. I've been a long time FreeNAS user, and one of my systems has rode the spinning disks into the ground (7 year uptime on most of them). So many errors in SMART on them, and things are starting to get pretty bad on the SMART stats.

I don't need a lot of space, but thought that an all-flash pool would give me a nice throughput for the usage on here (VMware ESXi storage), to match my Enterprise all flash systems I use at work. And perhaps a nice upgrade would be to get a used PowerEdge720 and fill up some bays with consumer grade 1TB SSD drives since they seem to be at a nice price point. So I started swapping out some of the worst pool disks with them as the start of a migration, and as a test to make sure FreeNAS likes the disks I got before I go all in on a build.

First I used SanDisk SSD Plus 1TB drives in a couple slots.

But then I started getting Checksum errors on both of the SSD disks. Not a lot, but enough to trigger Degraded mode.

NAME STATE READ WRITE CKSUM
vault DEGRADED 0 0 0
raidz2-0 DEGRADED 0 0 0
gptid/cafa41db-39a2-11e6-bd68-0010180687e5 ONLINE 0 0 0
gptid/1ba4b501-8e9c-11e9-9a79-000af71e9738 DEGRADED 0 0 317 too many errors

Then I swapped out the SanDisk SSD Plus 1TB disks with a Samsung SSD 860 EVO 1TB. But it too is getting checksum errors and also degraded the pool.

So, has anybody done an all-flash pool with either of these drives? Maybe its because they are such a different speed with the other pool disks until I can get them all swapped out and resilvered? Maybe it is the older SuperMicro 16-bay case I have them in now with the E1 backplane?
I didn't find a lot of feedback in the forums for an all-flash build.

Thanks.

 

[PhiloEpisteme]

So I started swapping out some of the worst pool disks with them as the start of a migration, and as a test to make sure FreeNAS likes the disks I got before I go all in on a build.

Did you do any tests or anything on the drives with your system before you replaced your in-use, spinning disks with them?

So many errors in SMART on them

What kind of errors?

So, has anybody done an all-flash pool with either of these drives?

It sounds like you had issues before you even got to an all flash pool, right? It sounds like you had vdevs composed of both flash and spinning disks?

Did you run a scrub on your pool before you started replacing disks?

 

[sretalla]

Could be an issue with the cable or SATA port... have you checked those?