Constant disk activity

[elfidge]

No luck with emboardmand - that must be a custom package from Asustor.
I can't get iotop to run as it doesn't recognise the kernel and/or have some prereqs installed.

My thinking is it's something reading temperatures and then writing this to a log. But I have no clue, I'll try support myself.

[elfidge]

In fact I just noticed in System Information in the GUI, that http logging was enabled and turned this off. So far I've seen no LEDs change or flicker since - can you change this and see if it seems to fix it?

[nero78]

In fact I just noticed in System Information in the GUI, that http logging was enabled and turned this off. So far I've seen no LEDs change or flicker since - can you change this and see if it seems to fix it?

Changed the HTTP logging. No change. My disks are being grinded every 7-8 seconds still...

Going to make a new ticket for this. Kinda regret buying this NAS now...

[elfidge]

Well I've tried pretty much every vendor and Asustor is my preferred choice. They have issues, but far fewer than others out there.

I've been playing again this morning and with a clean boot, ADM 2.60R4R1 on 6204T, all apps stopped, all logging off, all services stopped - I still see this sequential activity from left to right, so I'm convinced it's RAID syncing. Even running "iostat -dx /dev/md1 1" it's almost impossible to see anything - it constantly shows 0s throughout:

Device: rrqm/s wrqm/s r/s w/s rkB/s wkB/s avgrq-sz avgqu-sz await r_await w_await svctm %util
md1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Device: rrqm/s wrqm/s r/s w/s rkB/s wkB/s avgrq-sz avgqu-sz await r_await w_await svctm %util
md1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Device: rrqm/s wrqm/s r/s w/s rkB/s wkB/s avgrq-sz avgqu-sz await r_await w_await svctm %util
md1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Though the point is even having this running may cause a write itself, unless it is completely loaded in to RAM.

Previously I had a 302T, with two magnetic disks in R1. This would spin down when not in use, but I think R5 calls for checksums to be synced and checked across the array at all times to ensure consistency. If you put a R5 array all in to standby, then they'd have to spin up at exactly the same time and check consistency, or potentially even rebuild, so I'm not sure it's possible for R5 to go to "sleep". At least I've never had anything like that in years of production servers, they'll go in to low power mode, but always are active.

That's just my punt, so see what they say and report back please, would be useful to know.

[elfidge]

Oh the joys of google. See here:

http://geovoyagers.de/earth/linux/raid5-in-standby

Looks like it can be done, and hdparm is part of the Linux build Asustor are using. May be worth mentioning that to them? I wonder what their GUI actually uses in the background to run disk maintenance, I'm guessing that app they asked you to use previously, but this seems like a better option maybe, perhaps.

[nero78]

I have 4 WD RED's in R5 and i can get them to spin down (called disk hibernation in the ADM). So the NAS can "ignore" the R5 consistency check when disks are in hibernation mode?!

Have you tried the NETSTAT app? It's new and provide a TON of information. I have no clue how to read all that data...

http://netdata.firehol.org/

[orion]

I think there are basically 2 cases:

• Disk power mode status check. It is a command to get HDD power status without physical media access. I use "hdparm -C /dev/sdx" to check HDD power mode status, and I can see the green LED flashing once if I execute the command. This command should be fine (although we'll see the green LED flashing periodically). I can see this LED flashing before HDD going to spin down mode. It means that this command will not block HDD into spin down mode.
• Physical disk access. That should be caused by certain process. It will prevent HDD going to spin down mode.

You might need to check what's your case. I installed many Apps in my NAS, so I put OS related stuffs in one HDD and user data to other HDDs (RAID-5). My system HDD won't go to spin down mode. However my data HDDs will go to spin down mode as expected.

[elfidge]

Can you confirm what the output of that hdparm is please? Every single mounted volume has a status: unknown in my NAS.

As for separating out "OS" from "Apps", well this may be an option for an 8 or 10-bay NAS, but for anything 2 or 4-bay, of course the option is to go R1 or R5, so it's simply not an option. Either way as I've stated before, this did previously work on the AS302T with two spinning disks in R1 (where of course the OS would have been mounted), and these did spin down.

There's no question this is a code change between versions of ADM.

[orion]

Can you confirm what the output of that hdparm is please? Every single mounted volume has a status: unknown in my NAS.

As for separating out "OS" from "Apps", well this may be an option for an 8 or 10-bay NAS, but for anything 2 or 4-bay, of course the option is to go R1 or R5, so it's simply not an option. Either way as I've stated before, this did previously work on the AS302T with two spinning disks in R1 (where of course the OS would have been mounted), and these did spin down.

There's no question this is a code change between versions of ADM.

Here is the output (HDD is active):

Code: Select all

root@AS-304T-09D3:/volume1/.@root # hdparm -C /dev/sdb

/dev/sdb:
 drive state is:  active/idle

After HDD entering spin down mode:

Code: Select all

root@AS-304T-09D3:/volume1/.@root # hdparm -C /dev/sdb

/dev/sdb:
 drive state is:  standby

As I remembered, there is another discussion topic in the forum replied by an asustor guy. They are developing a tool to help us know which processes being active to access HDDs. I think that should be definitely the tool we want.
Anyway, my AS-304T configures as one single volume and one RAID5 volume. I'm pretty happy with this configuration. However the tool can help me to see which processes are active on system volume.

[John@AST]

Greetings,

Thanks for your patience. We are now testing the beta version of the diagnostic tool. The tool will show all of processes that access HDDs during certain period. You can then decide if you want to stop those processes to prevent constant disk activity.

The tool is embedded into ADM (under Settings -> Hardware -> Energy Control -> Diagnostic Tool). If you are interested to test this beta ADM version, we are happy to provide a link for download. Please leave your model name here.

NOTE: It's a beta version. So, please backup your data first.

Thanks.