N8800 unresponsive after 3.00.08 firmware update

[DoctorSAN]

Aaargh I just knew I'd been extra lucky until now with my N8800 boxes after reading peoples experiences here...

Just applied v3.00.08 to an N8800 which was previously at v2.01.09. Everything appeared to go well, however after the reboot, no access to the admin GUI can be gained via either of the LAN interfaces, and the front panel display indicates one drive is offline with no RAID arrays configured.

The 'offline' drive appears to be responding again after a power off restart, yet there is still no access to the GUI (both network ports will repond to a ping, but no http or https connections) and no sign of the two RAID5 arrays which previously existed.

I guess a support ticket with Thecus is going to be needed, but that sounds slow from what I read here, so was wondering if anybody else has already been through this..

As a point of interest, the same firmware was applied yesterday to another N8800 on the same site, which was previously using v3.00.07 - this one worked. At least it *seems* to have worked, but now I'm suspicious and don't really want to tamper with it since it's the only one carrying any data.

Though I get the feeling that the following info is not so relevant to such a fundamental defect, configuration for the N8800 which failed is 8x Seagate 1.5TB drives in two RAID5 arrays using XFS and accessed only via SAMBA shares. No modules or extras not supplied by the factory. Configuration for the N8800 which updated as expected is 8x Seagate 2TB drives in one RAID6 array using ZFS and accessed only via SAMBA shares. Again, no extras.

Suggestions would be greatly appreciated at this point..

[Karniv4]

Try using HTTPS rather than HTTP. worked for me after some digging.

(Although reading through youre post properly shows you mentioned trying https already - doh, ill shut up now)

Andy.

[DoctorSAN]

Nah, bring it on.. I'm desperate for anybody to throw me a bone. even looking closely at that factory_test.sh option mentioned on the other 3.00.08 thread.

How on earth do these firmware revisions ever escape the factory?? Are we humble users nothing more than stupid beta testers who paid for the privilege?

It really makes it difficult for a consultant to place hand on heart and specify these products to a client, when the 'flakiness' factor is undeniable. "Your ~16TB of data might or might not be available tomorrow sir". Pfft. Powervault here we come.





Simply because the pre-prepared icon is so apt for this discussion..

[Karniv4]

Agreed, I've put this box forward as a solution to a problem, and Thecus's methods make me nothing but nervious. I've still not had all the problems other folks are having though, and i'm using the main stream business side functions (AD, iSCSI, SMB shares,Webserver module) to destruction.

On the other hand, the difference in cost £3k vs £10-£15 of a HP solution, If i had the budget, 2 additional redundant boxes for a total of £9k would make me feel alot happier, and still be cheaper than a HP solution.

If only they could get there act together, sort out a consistent web site and proper QA....They could nail this market. If this turns out to be manufacturing variances though, they need to be aware they are doing more harm than good, or stick to the consumer goods.

Sorry Doc, nothing in this post is of help to you! Good luck!

Andy.

[DoctorSAN]

On the other hand, the difference in cost £3k vs £10-£15 of a HP solution, If i had the budget, 2 additional redundant boxes for a total of £9k would make me feel alot happier, and still be cheaper than a HP solution.

Did you mean you're using some sort of elastic storage pool which would tolerate units being offline for as long as is needed, or in terms of keeping spare units on the shelf and swapping out the drives if one dies or chokes on an 'update'?

I used to think the playing field was tilted toward the Thecus gear too, and maybe it is when compared to HP stuff though I'm not familiar with their product range (perceived bang-for-buck is how we originally ended up deploying the N8800 boxes), but another unit we use is the Dell Powervault MD3000i, which at last look in it's most basic config of two SAS drives for the OS, is a shade over $4k USD. That's an iSCSI unit with 15 drive bays and plenty of processing horsepower.

Even before reliability is factored in, the gap seems to be shrinking or non-existant for some purposes.

[DoctorSAN]

Ok, a tiny bit more to pass on.

The unresponsive unit had it's drives disconnected as a precaution, and was reverted to version 2.01.09 via the factory_test.sh script kindly provided by Peter at http://www.futterknecht.at/index.php...d=60&Itemid=78 after which it responded normally.

From there it was upgraded following the same path as the other unit here which didn't have a problem - 2.01.09 -> 3.00.07 -> 3.00.08 and this time was still responsive while running 3.00.08.
No idea if this upgrade path has any bearing, but I wasn't going to invest the time trying it 10 different ways to figure out what's failing until we see some Thecus cheques for beta testing services.

Drives were reconnected, the RAID was recognised, and after replacing the missing users and small details, the arrays could be seen once again.

Unfortunately, the motivation for all of this pain was to see if Nsync between N8800 boxes was now vaguely reliable, but sadly nothing has changed. N8800 3.00.08 syncing with another N8800 3.00.08 still aborts consistently on certain files early on in the job, without regard for file name, size, type or permissions (so it seems). Also tried the new Rsync alternative, but this never gets off the ground - reporting 'out of space' after a short wait, even though the test folder to be copied is in the order of gigabytes and the target folder has >10TB free.

Reading the release notes for 3.00.08 you might come away with the impression that ZFS does not play well with several of the functions surrounding Nsync and FTP, so the target array was rebuilt as one large XFS array, but Nsync tasks continue to fail.

Oh a quick note. Seems that the power management option for the drives now pretty much ignores you in 3.00.08 regardless of what you set it to. Both units (different model drives in each) spin down the drives after maybe 10-20 min of inactivity. Haven't tried every setting, but 240 and 300 sec are being ignored at least. Trying the 'never' option at the moment. Two steps forward and three backward it seems..

[Ruediger]

Hello DoctorSAN, Karniv4

Agreed, I've put this box forward as a solution to a problem, and Thecus's methods make me nothing but nervious.

Same here. We suffered a couple of time from similar problems (we posted a couple of threads here). We are at the point, we want to trash these devices through the window.
The quality dropped to a level, where they disqualify themselves to be used professionally. On every reboot we are afraid, whether it is working or not. Every FW update is a thriller.
I think Thecus lost its focus on reliable BASIC functions. They were forced to develop colorful and shiny GUIs, since competitors did. Thecus did not realize that development of professional NAS/firmware is completely different from these home toys.

[DoctorSAN]

Exactly my thoughts. I don't closely follow the habits of Thecus the company, but was amazed to recently see the effort put in to the fancy Ajax type interface, when basic functions like Nsync are badly broken requiring homebrew user workarounds, and appear to have been for some time. I mean, even in a HOME environment, how long do we spend in the user interface Thecus? A competing product line (Infortrend) doesn't even bother with a GUI - it's all VT100 and ANSI character menus and I doubt anyone is complaining that it isn't pretty enough.

If the unit is halfway reliable, time spent in the GUI ought to be in the order of 'minutes' per year, no? I don't wish to be made comfortable in the GUI, I want to set it and forget it.

Give me something that is ugly but works anyday over a Qnap-lookalike coat of paint.

[Ruediger]

Give me something that is ugly but works anyday over a Qnap-lookalike coat of paint.

100% ACK. Reliability comes first. Then features and last: the GUI. Thecus has turned round the order. I guess, they are convinced they could not sell units without a shiny GUI. QNAP is building up pressure.
They should have a look on the professional devices (e.g. NetAPP) and copy them and not their direct competitors.

I am convinced that they will lose ground on the NAS market, if they do not solve their catastrophic quality problems.

For me, the value of a NAS is defined by its reliability, not by the number colors used in the GUI.

[Karniv4]

Did you mean you're using some sort of elastic storage pool which would tolerate units being offline for as long as is needed, or in terms of keeping spare units on the shelf and swapping out the drives if one dies or chokes on an 'update'?

I wasn't thinking about FW updates, more day to day security/redundancy. You typically don't expect a piece of hardware to be unrecoverable due to a FW update, yet here we are, with quite a few complaints and no recovery method.

In all fairness, unless there's a business case for applying a new firmware, I would tend to avoid it. Both time's i've updated this N8800SAS Nas have been due to a request from Thecus support for the issues i've had (So far 2x freezes - Seems to be fixed from 3.00.07a onward, and various Nsync issues - which it's too early to say if they've been fixed by 3.00.08).
Unless there is a compelling reason this NAS will stay at this FW version for the rest of it's life, especially as the number of random failures seems to be fairly high. I'm beyond the point where I can call it a development box!

Andy.

[DoctorSAN]

I wasn't thinking about FW updates, more day to day security/redundancy. You typically don't expect a piece of hardware to be unrecoverable due to a FW update, yet here we are, with quite a few complaints and no recovery method.

Understood, but I was curious about how you are using this box (or boxes), that could tolerate the failure of an entire NAS while you substituted one of the other cheaper Thecus units, purchased for less than the total cost of the abovementioned HP box.

Guessing this is some sort of super-redundant arrangement?