D5-300C Device Resets

USB RAID storage for photographers, music producers, movie collectors
User avatar
TMRyan
TerraMaster Team
Posts: 817
Joined: 01 Dec 2020, 11:50

Re: D5-300C Device Resets

Post by TMRyan »

Sorry,
I thought you were running CentOS 7.9.2009 in a virtual machine of Raspberry Pi4.
Can D5-300C be used normally on Windows?
Is it possible to let our engineer check it remotely? You can send an email to our tech team :viewtopic.php?f=75&t=608
To contact our team, please send email to following addresses, remember to replace (at) with @
Technical team: support(at)terra-master.com (for technical support)
Service team: service(at)terra-master.com (for purchasing, return, replacement)
User avatar
pgreenland
Posts: 7
Joined: 05 Jun 2021, 07:49

Re: D5-300C Device Resets

Post by pgreenland »

Hey t3d,

Did you get anywhere with this?

I'm seeing very similar if not exactly the same issues with a pair of D5-300C's connected to a linux machine.

Using linux's UAS driver, they appear to work fine under low load, but with lots of IO you begin seeing resets and eventually the drives become unresponsive.

With UAS disabled and the older usb-storage driver in use the resets appear occasionally but the drives remain functional.

Not entirely sure I trust them with my work though.

I bought two of these units to replace an old NAS, so far they're destined to go back to Amazon to be replaced with QNAS units.

Be a bit of a shame if they do have to go back, other than not full-filling their primary function they look good and appear reasonably well built.

Let me know,

Thanks,

Phil
User avatar
t3d
Posts: 12
Joined: 20 Mar 2021, 22:20

Re: D5-300C Device Resets

Post by t3d »

Hi Phil,

Sadly, no. I’ve not had a second cataclysm, but usage is low and I do still see the errors when things are happening on the drives.

Ted
User avatar
t3d
Posts: 12
Joined: 20 Mar 2021, 22:20

Re: D5-300C Device Resets

Post by t3d »

I’m very curious to know which model qnap you’re looking at. I’ve not been able to find anything competitively priced with 5 bays.
User avatar
pgreenland
Posts: 7
Joined: 05 Jun 2021, 07:49

Re: D5-300C Device Resets

Post by pgreenland »

Hey Ted,

Not good to hear....but good in a way. I'm glad to see someone having exactly the same sorts of issues with the enclosure as me. I've been cursing my machine's USB implementation and chasing all sorts of potentially unrelated USB and IOMMU errors appearing in the kernel log around the same time as the disk errors for the last week or so.

Ordinarily my usage is low, the enclosures were initially going to contain some large disks for backup and archive. With a view to absorbing the disks from my old NAS once I trust it. However any time I stress it by performing a backup, or offloading a load of gopro footage it starts barfing out USB errors.

I've just ordered a QNAP TL-D800C with 8-bays to replace the 2 x D5-300C's. I paid £180 for each of the D5's on Amazon. With the QNAP being slightly more at £407 from Box (slightly more as in £50 more than the pair of terramasters). Admittedly I've lost two drive bays, but only needed 8 to begin with.

Price wise I wasn't able to find a 5-bay unit that beats the terramaster. It looks good (be nicer in black), feels reasonably well build and has a sane interface. Even performs fairly well when its not causing the host to report errors. Would be very happy with if it wasn't for the USB issues.

Just hoping I don't experience the same thing with the QNAP :roll:

Thanks,

Phil
User avatar
t3d
Posts: 12
Joined: 20 Mar 2021, 22:20

Re: D5-300C Device Resets

Post by t3d »

Let me know how you get on with the QNAP. If it’s smooth sailing I may need to upgrade. I agree with you this unit has many fine qualities. But fundamentally a DAS must be a safe space for my data. I wish Terramasters support wasn’t so abysmal. Maybe we could get it fixed.
User avatar
pgreenland
Posts: 7
Joined: 05 Jun 2021, 07:49

Re: D5-300C Device Resets

Post by pgreenland »

Hey Ted,

The TL-D800C that I ordered arrived a few days ago.

Have loaded it up with a couple of 10TB IronWolf HDDs and a pair of Micron 5300 Enterprise SSDs.

Construction wise, its generally pretty good, seems to deaden the noise of the IronWolf drives more than the Terramasters. The only bit I've not been impressed with is the screwless drive bays. They have a very weak feeling clip on strip down either side, which holds pins, lining up with the drive screw holes. I've carefully removed these and screwed the drives in from below, although you might be able to screw though the rubber pumpers, similar to the TerraMaster arrangement.

Each drive bay has a status LED above it, which appears to accurately reflect activity.

User variable fan speed via a switch on the back, virtually silent when on auto (drives are around 35 degress, which given the 27 degree ambient today doesn't seem too bad).

10GBps USB 3 Gen 2 rather than the 5GBps Gen 1 - similar arrangement of a hub and various USB to SATA bridges - linux reports them all operating at 10000mbps with the uas driver in use for all disks. Only slight limitation is a 1 meter cable length to operate at these speeds, although thats seemingly a universal USB recommendation, longer cables may work, but I've not yet wanted to tempt fate.

I'm seeing around 600MB/s when writing to both SSDs at once, and 400MB/S when writing to a single one. It feels a little slower than I was expecting given the higher interface speed, so may warrant some further investigation at some point. Given that I haven't measured the performance of the SSDs when connected directly over SATA, it's hard to make a direct comparison.

The main thing though, over 40TB written now, filled the IronWolves twice and the SSDs a few times too - not a single USB or SATA related message in the kernel log after having cleared it post startup several days ago. Have been querying via smartctl too, for health and temperature monitoring, again no issues.

Quite happy with this as a setup so far, will be migrating my old NAS drives into at the weekend, returns booked for the terramasters.

Good luck in getting your setup stable!

Thanks,

Phil
User avatar
t3d
Posts: 12
Joined: 20 Mar 2021, 22:20

Re: D5-300C Device Resets

Post by t3d »

Thanks for the follow-up, Phil. That answers the core question. :) The issue is the Terramaster device. Time to bin it and replace it with something that does the job without risk to my data!
User avatar
pgreenland
Posts: 7
Joined: 05 Jun 2021, 07:49

Re: D5-300C Device Resets

Post by pgreenland »

Hey Ted,

Me again :-(,

Out of interest, are you running an AMD or Intel processor?

I've just loaded a few more disks into this new enclosure.....and....dun dun dun.....the UAS errors are back under heavy load. If it wasn't so heavy and full of hard disks I've have chucked the thing out the window at this point. Seems USB may have to be limited to mice and mug warmers.

The errors with this new unit seem to be recoverable at least. The kernel appears to wait for a timeout (possibly a disk or SCSI timeout, not entirely sure which yet). Resets the device and re-issues the commands, which then complete. Exactly the same pattern as the terramaster unit though.

I took the brave step to leave Intel behind and jump on the Ryzen bandwagon, think my last AMD processor was something like a K6 :-P. After which I started reading the reports of USB issues, which I believe still affect the 4800U I've got in this mini-pc.

If it turns out the problems are all AMD USB bug related an apology to terramaster may be in order. Followed by a ceremonial firing of AMD's garbage into the sun as an offering to intel, hoping they'll take me back. :lol:

Thanks,

Phil
User avatar
t3d
Posts: 12
Joined: 20 Mar 2021, 22:20

Re: D5-300C Device Resets

Post by t3d »

Hi Phil,

I’m actually on a Raspberry Pi 4. So neither intel nor AMD. To the best of my ability to discern this issue is related to the firmware used on the asmedia usb chipset. Newer revisions of the chipset don’t have this problem. In theory, Terra-master could fix it with an update — but that may depend on how they handled the usb firmware in their implementation.
Locked