encrypted volume creation, safely initialized -> how long?

[Mr_N]

Just wondering if anyone can give me an estimate of the time it will take to create my encrypted volume with the initialize safely option 6x4TB RaidZ2 (~14TB)

 

[cyberjock]

I'd say 2-3 hours, perhaps longer. It really depends on your disks. If you go to the CLI and type gstat you will see the writing to the disks. When it's done it will setup your zpool and the busy indicator will sit at zero. Honestly, I'd just let it run overnight and check it in the morning. :)

 

[Mr_N]

oh cool i thought it would be much longer :)

so typing gstat in the cli wont affect what its doing?

 

[cyberjock]

Nope.

The disk speed and random number generator speed will determine how long it takes. If you have a 100GB drive it might take only 20-30 mins. The most accurate answer you can get is "it'll be done when it's done".

 

[Mr_N]

I think maybe it'll take longer than you think ;)

 

[Mr_N]

so when its done the volume will just appear under active volumes?

the gstat still shows writing to the drives, but the CLI is now displaying lines saying:

volume name kernel: pid 6739 (dd), uid 0 inumber 9282 on /var: filesystem full

<edit> seems theres 6 tmp files in tmp dir all 21MB each taking up the space

 

[Mr_N]

So after the system finally stopped writing encrypted data to the drives, the volume never showed up so i just remade it without ticking 'safely initialize' this time and worked immediately, and now its all working as expected.

 

[HolyK]

I started "Initialize safely" on my two 2TB REDs (which are in mirror) yesterday and when i was going sleep it was running for 12 hours. I did not had time to check the status this morning but i suppose it finished, or at least i hope :D

 

[a_b]

I had the same problem just now.
Tried to secure initialize 10 x 4TB hard disks for Raidz2 as encrypted volume, and after about 2 days the messages came that filesystem is full in /tmp/....
I do not know what or why sent this, but this feature is clearly not working properly as it is implemented.

Some kind of progress report would not hurt either while this is running (was checking the system processes every few hours and saw the 10 x dd instances going, maybe that could be used somehow for a progress report).
In any case, did the hard drives fully initialize with random data as it was supposed to do or the process has stopped at 10% or 80%? Who knows?

Any way to check this?

 

[cyberjock]

I just always do it manually using dd. There's no real way to know if it finished or not. Even doing sector reads, unless you happen to read a sector that has clear plaintext you're not likely to be able to distinguish the difference between what came from /dev/random and what was data that was on the disk in the past.

 

[dlavigne]

If it is filling /tmp and bailing out, that is a bug which needs to be reported. Please create one at bugs.freenas.org and post the issue number here.

 

[ghackett]

Thought it might be worthwhile to note that I've been running the Initialize Safely bit for about 48 hours now on 8x4TB Red drives (RAIDZ2) backed by a 4-core xeon E3-1220v3. `top` still shows 8 instances of dd runnning each owning ~44% of a core, along with python2.7 owning the leftover 50-55%. `gstat` shows my disks only writing at between 10,000 ~ 13,000 kBps, which is kind of a bummer and I think means the process is going to take ~5 days (if I've done the math right). I'm curious if the ops here would see that as a bug or standard for this task. Unfortunately, my server doesn't have a video card so I can't see any console output without shutting it down first and putting one in.