Question on Syncronization

[Mrantala]

Hello -

Sorry if I'm posting this in the wrong section, but new here, and new with Asustor in general. I recently purchased the AS4004T for use at home. To start with I also purchased to brand new Seagate Exos 14tb hard drives. So, this week, I just wanted to be sure the unit was good, so I set it up, went through with the quick setup, which configured it to a Raid1 with a total of about 12.5tb free storage space. THAT'S IT. I didn't do any more than that, I didn't even set up the network path or anything to access the storage space. So what ever was put onto the drives was put on by the ASUSTOR setup - which can not be that much at all.

I noticed however, that the hard drive activity lights were going crazy with activity. Since I was sitting in the office near the box, I especially noticed it. Even after about an hour, they were still going crazy. I was done for the day, and didn't want the drives running like that, so I shut it down. Came back to this today turned it on, and the same thing was going on. So after some digging, it appears that the NAS was synchronizing the drives in the RAID1 volume. OK.. but when I looked at the details, it's a LONG PROCESS ... showing an estimated 20 hours !! before being completed ? (see screenshot attached).

How can this be for a volume that has basically little to no data on it ? What could it possibly be "synchronizing" beyond the initial files loaded by ASUSTOR during setup ? What is it "transferring" in that 12.4 TB of blank space on the disks ? What am I missing here ? I appreciate any insights into this ? For right now, I've deleted the RAID1, and set it to RAID0 for now.

Thanks in advance !!

Why is the synchronizing of this volume (which is basically empty) taking 20 plus hours to complete ?

WhySoLongToSync.JPG (41 KiB) Viewed 3798 times

[Nazar78]

That's quite normal, depending on the size of the array and background activities but you're still able to use the NAS normally. Advise you not to use raid 0 unless you don't mind losing the data when it breaks. Google more about raid.

[Mrantala]

Hey Nazar78....thank you for your post back on this. So, yeah, my concern was this 20+ hour sync process and why that would be on a basically blank RAID1 Volume. So, I did a test with a much smaller RAID1 setup. Took out both 14TB Drives and replaced with 2 drives, for a max of 1TB Volume. The system did the same process with this setup as well, but the sync only took 5 hours. Then after that, the drive activity was as expected and were pretty quiet.

So I'm assuming that the 20+ hours sync of the 14TB RAID1 where the drives are running non-stop is done during that initial setup and probably won't see that kind of activity unless of course a drive fails, or I would decide to add drives to the NAS for more storage, etc. Good to know the drives won't be running 24/365! So I'll put the 2 14TBs back in today, set them as RAID1 and let them rip for a day!

Thanks ! - Matt

[Nazar78]

NP yup you're right. It also depends on raid type. My raid 10 4x10tb (typically a mirror and stripe) took about 11 hrs to restore one failed disk.

You could however speed up this process but not recommended when the nas is in use. Google about below.

Code: Select all

echo 500000 > /proc/sys/dev/raid/speed_limit_max
echo 500000 > /proc/sys/dev/raid/speed_limit_min

[Mrantala]

Hey - that's good stuff ! I checked it out, and I'll probably set those 2 values before I try the initial sync of the 14tb RAID1 and see if we can trim a little time off of those 20 hours. It's not that I have to find a way to trim time off... now it's just for my own personal curiosity. You seem pretty well versed in this NAS arena, so I'm sure you probably know most of this already but, doesn't hurt to pass it along. Researching those two lines, I found this site with that tip, and a few others that you might try as well....

https://www.cyberciti.biz/tips/linux-ra ... speed.html

Much appreciated ! -Matt

[Nazar78]

Yup2 great to share. There's another tip I personally stumbled upon but can't find any exact references on the net except why is it slow? The resync is slightly faster if you set the io scheduler other than CFQ. Deadline works well, avoid NOOP unless the disks are headless.

Usually I'll just set these speed limits when doing recovery as it's not permanent and will revert on reboot unless of course I want to make it stay. The bitmap I'll pass because I need the random write performance. From the link you've shared, some won't apply to ADM i.e. the ones before tip #1 unless you manually build it.

IMPORTANT
Most are also non-persistent and safe to use except "Tip #4: Disable NCQ on all disks" albeit non-persistent, on ADM you won't be able to set it to on for external drives as the hub can't determine the NCQ capabilities but you will get some md sync errors if you were to set it to off for internal drives. Edited: found the error - Frozen Emask 0x10 (ATA bus error)