I can't stress enough how important it is to a) open support tickets AND b) point them to this thread here. They won't listen and run you through first level support questionaires (have you tried rebooting?) for weeks otherwise. Point them here.
Unfortunately my support ticket was terminated because Crucial MX300 is not on the supported list. Apparently you guys who are experiencing the same issue with compatible drives could have a different root cause.
They suggested I purchase a 'compatible' SSD, which to me is a somewhat expensive experiment since the root cause is unknown and given the similar symptoms there is a good chance it will do the same.
For what it's worth, I have definitely narrowed it down to the caching. Setting the two SSDs up as individual volumes, I could not reproduce any slowdown with multiple operations between them via the console. But setting one SSD as a read cache (!?) and copying from HDD to the other SSD, I saw the slowdown after 100-odd GB.
I could live with it if it would drop to HDD-like speeds. But it goes well below that, 15-20MB/s is just not fun.
Wow... what a shabby excuse from them... the drive is not compatible. That's one way of saying to f§%$ off. QNAP support keeps failing, as usual. Did you point them to this thread here?
I've received my support email this morning from tech support, basically saying windows 10 does have transfer issues and also asking for log files from my Qnap to be sent over.
I'm at work at the moment but will reply when I get in tonight and reiterate that the issue is with SSD raid cache and is resolved only when disabling the SSD cache acceleration, nothing to do with windows 10 as the speed is at max /without SSD caching, along with sending the logs they asked for AND pointing them to this thread.
cryptochrome wrote:Wow... what a shabby excuse from them... the drive is not compatible. That's one way of saying to f§%$ off. QNAP support keeps failing, as usual. Did you point them to this thread here?
Yup, pointed this thread out in my initial ticket. Apparently people with supported drives who are experiencing the same issue "could have a different cause", therefore my not-on-the-list SSD has been labelled as "incompatible" without understanding what the cause is.
I opted for the MX300 because, as it was a new model, I didn't know how long it would take to be tested, if at all. Not an unusual dilemma with new models.
pokrakam wrote:Unfortunately my support ticket was terminated because Crucial MX300 is not on the supported list. Apparently you guys who are experiencing the same issue with compatible drives could have a different root cause.
They suggested I purchase a 'compatible' SSD, which to me is a somewhat expensive experiment since the root cause is unknown and given the similar symptoms there is a good chance it will do the same.
For what it's worth, I have definitely narrowed it down to the caching. Setting the two SSDs up as individual volumes, I could not reproduce any slowdown with multiple operations between them via the console. But setting one SSD as a read cache (!?) and copying from HDD to the other SSD, I saw the slowdown after 100-odd GB.
I could live with it if it would drop to HDD-like speeds. But it goes well below that, 15-20MB/s is just not fun.
That is an incorrect statement by support. Mx300 are on the supported list for tvs-x73 family.
Sent from my iPad using Tapatalk
Paul
Model:TS-877-1600FW:4.5.3.x QTS (SSD): [RAID-1] 2 x 1TB WD Blue m.2's Data (HDD): [RAID-5] 6 x 3TB HGST DeskStar VMs (SSD): [RAID-1] 2 x1TB SK Hynix Gold Ext. (HDD): TR-004 [Raid-5] 4 x 4TB HGST Ultastor RAM: Kingston HyperX Fury 64GB DDR4-2666 UPS: CP AVR1350
The compatibility list only includes the M.2. As it's just a different form factor, I figured a similar device would have a good chance of success. I'm still not convinced there's any compatibility problem.
Of course there is no compatibility problem. They are just using that statement as an excuse to not have to deal with you and a problem they have no clue on solving.
pokrakam wrote:The compatibility list only includes the M.2. As it's just a different form factor, I figured a similar device would have a good chance of success. I'm still not convinced there's any compatibility problem.
They are both SATA interfaces. They both use the same FW files from Crucial (which would imply same/similar Marvell SSD controller). I would challenge them to show you the difference.
Tell the first level tech that you want the issue escalated to RD (Taiwan).
Paul
Model:TS-877-1600FW:4.5.3.x QTS (SSD): [RAID-1] 2 x 1TB WD Blue m.2's Data (HDD): [RAID-5] 6 x 3TB HGST DeskStar VMs (SSD): [RAID-1] 2 x1TB SK Hynix Gold Ext. (HDD): TR-004 [Raid-5] 4 x 4TB HGST Ultastor RAM: Kingston HyperX Fury 64GB DDR4-2666 UPS: CP AVR1350
Trexx wrote:Tell the first level tech that you want the issue escalated to RD (Taiwan).
RD are the ones who have been on the case, the exact response I got:
Support wrote:Its hard for us to narrow down when it comes to debugging. Even if we want to reproduce the issue and analyze it which is not possible to do it if you the incompatible SSD drives.
Issues may sound common but root-cause could be different.
I will just wait until QTS 4.3 with a new kernel and/or for someone with this issue on a compatible SDD to get their ticket resolved, which should make my incompatible SSD compatible again
In the latest QNAP Diagnostics tool, there is an SSD dump option (at least under 4.3) that generates a file giving some information from Flashcache.
Have they had you do that at all?
You are supposed to run the SSD Dump while the SSD is getting heavily hit. Might try doing your transfer test where you see the speed drop. If you know roughly when the speed drop will happen, you could start the SSD Dump just prior the speed drop so it captures before/after.
Might be worth a shot.
Paul
Model:TS-877-1600FW:4.5.3.x QTS (SSD): [RAID-1] 2 x 1TB WD Blue m.2's Data (HDD): [RAID-5] 6 x 3TB HGST DeskStar VMs (SSD): [RAID-1] 2 x1TB SK Hynix Gold Ext. (HDD): TR-004 [Raid-5] 4 x 4TB HGST Ultastor RAM: Kingston HyperX Fury 64GB DDR4-2666 UPS: CP AVR1350
Hmm, they did use that and took the complete dump log, but not while it was struggling. I only just noticed the SSD option at the bottom now that you pointed it out, thanks!
Will give it a go but will take some time (4-6 hours to activate and synch the cache...).