QNAP firmware issues : Major file system bug

[umpa]

I'm just about to put the 500GB of data I had to move back on the NAS. I really hope its fixed it.

[grzlhmpf]

well,

here (639, 5.6TB RAID5 80% full) it is a lot better with 3.7.1... but still a lot slower than it is with an empty array. creating folders still takes a long time, but the NAS does not stall other connections meanwhile.

to me it seems they really improved the situation, but i wouldn't call it "fixed" (i expect FULL performance until the "disk full" message shows up)

Michael

P.S. PLEASE FIX THE STANDBY BUG!!!

[schumaku]

Michael, Ever heard about file sytem fragmentation?

Start with an empty storage volume, and just fill it up by wriging data in a single thread - the write performance will be almost constant until the file system runs full.

[grzlhmpf]

file system fragmentation at 80% running EXT4 with mostly large files (written in one stripe and then just read)? not really.

and i NEVER heard fragmentation makes directory creation slow.

i would still like a defrag tool though.

Regards,
Michael

[umpa]

Preliminary test seem to confirm that the NAS is now fixed. I have put all my data back on the device and it has not crapped out on me. Performance seems at least as good as when new or perhaps a little better. Its early days, and the benefits did not become apparent until a few hours after the update (I guessed the nas was busy) but now all seems AOK.

[umpa]

Nope I'm still having issues - it seems that after a restart the update broke bonjour services and AFP I can only connect via the GO>Connect to server and then not to AFP. I have confirmed this on 3 macs.

[schumaku]

Nope I'm still having issues - it seems that after a restart the update broke bonjour services and AFP I can only connect via the GO>Connect to server and then not to AFP. I have confirmed this on 3 macs.

And how is that related to the subjected topic here? No, in geneal Bonjour is not broken with the update. mDNSrepsonderposix was replaced by Avahi code. Consider to re-install the firmware using the manual process as shown here http://wiki.qnap.com/wiki/Manually_Updating_Firmware and in more detail in many forum posts. If there is an issue, please open a new forum thread please.

[AdrianW]

As I am still experiencing pauses I contacted Qnap (as described in this post: QNAP’s response to discussions about abnormal performance).

They provided a diagnostic qpkg which I installed - this produced a couple of reports that I sent to them yesterday. They've emailed me back today saying this:

Based on your log files, it seems that your issue is not caused by fragmentation, and the symptom is different from those users we've identified.

The plan now is to get a Qnap engineer to connect via TeamViewer next week to do a more in-depth check.

[umpa]

Nope I'm still having issues - it seems that after a restart the update broke bonjour services and AFP I can only connect via the GO>Connect to server and then not to AFP. I have confirmed this on 3 macs.

And how is that related to the subjected topic here? No, in geneal Bonjour is not broken with the update. mDNSrepsonderposix was replaced by Avahi code. Consider to re-install the firmware using the manual process as shown here http://wiki.qnap.com/wiki/Manually_Updating_Firmware and in more detail in many forum posts. If there is an issue, please open a new forum thread please.

Well, I think its related because as the drive fills up it becomes that slow it semi crashes, certain info pages refuse to come up, and the drive drops out from Mac and windows and the ONLY way to get it back is to use the web file manager and off load data - then all of a sudden its back to normal. I have also found out that given enough time it may recover on its own - but you need to leave it all night. I posted here because I believe that this is a related issue, and my symptoms may differ slightly due to the HUGE size of my array and the fact that the data is also HUGE (being raw uncompressed video) - so huge amounts of data is moved regular.

The fact that I only noticed what I thought was a Bonjour problem on 3.7 does not mean it was not there on 3.6, only that due to excessive testing on 3.7.1 did I reach that problem, and may very well be a side effect of the nas grinding to a halt rather than an actual bug with the service. Do you better understand the issues now ?

[2000wolf]

well,

here (639, 5.6TB RAID5 80% full) it is a lot better with 3.7.1... but still a lot slower than it is with an empty array. creating folders still takes a long time, but the NAS does not stall other connections meanwhile.

to me it seems they really improved the situation, but i wouldn't call it "fixed" (i expect FULL performance until the "disk full" message shows up)

Michael
...

I take your point - however I think such a behavior is quite acceptable for the low power processor used in our small NAS. Maybe someone with a i3 / i5 powered NAS can tell us more.

[moritzthecat]

looking to the recent remarks on fragmentation ...

As this hread was closed some time before with the remark on fragmentation issues I tried to find out about the fragmentation of my NAS TS-419p.
http://forum.qnap.com/viewtopic.php?f=25&t=59692
As fragmentation was confirmed to be on the very low end and I heard about issues concerning spin-down (ARM only?) I decided to wait for the update towards 3.7.1 and still use 3.6.1

Argumentation from QNAP still not very consistent to me.

Pls. let me know any feedback on the fragmentation measurement link posted above and about your experience concerning spin-down behaviour.

Thanx in advance.

[schumaku]

As fragmentation was confirmed to be on the very low end and I heard about issues concerning spin-down (ARM only?)

Intel only.

Still, if you have poor write perfromance, thw worth to go for 3.7.1. If the issue persists, please check here http://forum.qnap.com/viewtopic.php?f=160&t=60368

[AdrianW]

I've had a Qnap engineer connect to my TS-859 (via TeamViewer) for a number of hours both last night and tonight.

He's been trying to track down what's causing the long pauses that I'm experiencing. He took away a number of different log files.

He also tried out two different firmware versions, one was a variety of 3.7.1, the other was numbered 4.0.0, although this certainly wasn't the official next version.

I'm not sure, but I didn't really notice any difference between the versions - the NAS seemed to still be experiencing pauses.

All the tests were done within the NAS, so samba and networking didn't come into it.

[AdrianW]

I usually have my Qnap set to startup and shutdown each day, as it is only in use within a nine hour window, so I'd rather save the 15 hours of power each day.

But, as a test (suggest by the Qnap tech), over the weekend I left the NAS on for three days. I had an initial stall on the first day. But after that I experienced no further stalls during any copies which I did spread over the three days.

They're of the opinion that the stall is caused by the NAS "reading filesystem group bitmap". It seems to do this on the first copy after start up.

They're going to see if they can improve the initial load of the bitmap during system boot.

[broz]

I can confirm that the update has fixed the issue in my 12TB TS-659 Pro II Turbo NAS.

The problem used to occur with about 2.5TB remaining and would make the device unusable until space was cleared again. I currently have less than a TB remaining and everything is working at full speed.