8.0.3 RELEASE coming soon

[joeschmuck]

Good to hear you so absolute 100% sure about that.

100% sure. If the software were able to kill the drive then there would be a flaw in the drive. You might see if there is an issue with that drive from the manufacturer. Flash drives do have a limited life of write cycles for each bit but having the drive completely not recognizable generally indicates a complete failure vice just not being able to write, something more rare. During a software upgrade there is a lot of writing going on.

Must be one of the most useless and stupid answers i have ever seen on a forum.

It was done in fun, guess you're a bit too serious, I'm sure someone enjoyed it. I did give you sound advice with GParted and still recommend you give it a shot, even if you think it's unlikely, and it might not work but you have nothing to lose and everything to gain.

 

[angellayne]

well unless you did the upgrade during the leap second adjustment (compensates for the slowing earth rotation) at which point the software builds up a very small charge of gamma radiation and it has been know to slip out of the containment field and destroy nearby electronics or cause small animals to glow green for a few hours. I would suggest that either the USB stick was exposed to this rare phenomenon causing a hardware failure

I just laughed very hard! In fact I even fell out of my chair laughing, then I hit my glowing green dog. I found out that it wasn't a joke as I originally thought... :0)

ps. humor is a beautiful thing...

 

[gcooper]

... very small charge of gamma radiation and it has been know to slip out of the containment field and destroy nearby electronics or cause small animals to glow green for a few hours. ...

Wouldn't the USB stick just go Bruce Banner then :)?

In either case I would recommend replacing the USB stick. You haven't given any details on your system nor what brand/model USB stick you have. Do not use USB3 (just my recommendation, not the port, not the USB stick) because support is spotty. It will work for some, fail for others. Please provide your details on your hardware.

Either...

http://www.youtube.com/watch?v=rY0WxgSXdEE

... or your machine was one of the versions that actually worked with xhci. In which case you'll need to do:

1. Hit 6 when the FreeNAS bootloader pops up (drop into the bootloader console).
2. Enter in "load xhci" on the console.
3. Enter in "boot".
4. Add a loader as follows... variable name: xhci_load ; value: YES (see http://doc.freenas.org/index.php/Loaders ).

 

[gcooper]

Well... I don't have a machine to test with [at home or at work], so I'm going to continue on with the idea in mind that some machines will function better with AIO on, some off, and some will need the AIO read/write values adjusted. There are some other bits at work that can help performance with machines with higher core counts (kern.hz=100, pre-FreeBSD 9.x), so that might fix the interrupt issues, amongst other things.

8.0.3-RELEASE-p1 images will be posted near the end of the day, fixing the DHCP/DNS functional regression in 8.0.3-RELEASE, update ataidle from 2.60 to 2.72 based on the lead developer's feedback, fix the inadyn port on i386 permanently, fix xvid port build race condition issues, as well as the NFS without reboot bugfix noted by an iX customer.

This should close out the 8.0.3 RELEASE chapter so I can move on to testing out 8.2.0 for release in February or so.

 

[joeschmuck]

Sounds good, about releasing it, not the fact you don't have any computers to test the software on. And I'll bet you can't wait to test 8.2.0 :)

I agree, you simply cannot address every possible hardware situation and some people will have to tweak their systems a bit to optimize it. I for one thing the 8.0.3 version is working very well and will be the best "Release" version to date.

OT: I saw you created 8.2.0 branch the other day. If you create a thread titled maybe "8.2.0 Beta Testing" when you are ready for us to start testing, I'll jump in. I don't want to jump in too early while you're still making dozens of changes.

And thanks for all you do.

 

[gcooper]

Sounds good, about releasing it, not the fact you don't have any computers to test the software on.

I'm never satisfied unless I can test stuff on real hardware. I have a box at home, but it has a realtek (re) card in it and unfortunately (on FreeBSD 8.2), it detects the line speed as 100MBit instead of 1GBit (I know the card is capable of 1GBit on 9.x, and the cable is CAT5e..), so I can try doing a one-off test of 8-STABLE, but then it's not really a valid apples to apples comparison (and that's what I'm looking for).

And I'll bet you can't wait to test 8.2.0 :)

Indeed :).

I agree, you simply cannot address every possible hardware situation and some people will have to tweak their systems a bit to optimize it. I for one thing the 8.0.3 version is working very well and will be the best "Release" version to date.

Perhaps, but as long as things are generic enough, I'd like to have a good litmus test for consumer grade hardware as well as enterprise hardware to work with.

OT: I saw you created 8.2.0 branch the other day. If you create a thread titled maybe "8.2.0 Beta Testing" when you are ready for us to start testing, I'll jump in. I don't want to jump in too early while you're still making dozens of changes.

I moved it TBH; jpaetzel started it :D.

I've been hands off for the most part on trunk because I want others to finish their work before we get too deep into the release process instead of ending up with the divergence between 8.0.x and 8.2.0 like we have today, and to avoid double commits. The only thing I really want to solidify are service packs because I haven't seen them in action yet (even though some folks claimed they worked in the past).

And thanks for all you do.

Np. I try :).

 

[joeschmuck]

The only thing I really want to solidify are service packs because I haven't seen them in action yet (even though some folks claimed they worked in the past).

I didn't know any service packs existed. I'll find that thread later.

 

[gcooper]

I didn't know any service packs existed. I'll find that thread later.

Service packs in FreeNAS are the equivalent of the industry idea called hot patches. This is how 8.0.3 should have really been distributed for folks that were upgrading from 8.0.2, but I due to time constraints I couldn't do that.

 

[gcooper]

8.0.3-RELEASE*-p1 has been posted on SourceForge. Cheers!

 

[PDM]

8.0.3-RELEASE*-p1 has been posted on SourceForge. Cheers!

No community feedback on the new patched release? I was tempted to upgrade, from FreeNAS-8.0-RELEASE-amd64, this morning when I saw it freshly posted.

 

[ProtoSD]

No community feedback on the new patched release? I was tempted to upgrade, from FreeNAS-8.0-RELEASE-amd64, this morning when I saw it freshly posted.

I just barely been 12 hours, give it more time. People are generally slow to discover a new version has been posted. If you're using 8.0, than 8.03 should be a huge improvement. It's likely the best/most stable release so far!

 

[moraga695]

Just upgraded from 8.02 x64 release to 8.03 Release P1 using the x64.GUI_Upgrade.xz method with no problems noted. I have a question about the new CIFS AIO default. The Readme says "tune the AIO size from '1' to 4096". I assume this applies to the Minimum AIO size. What is suggested for the Maximum AIO size setting?

 

[joeschmuck]

Just upgraded from 8.02 x64 release to 8.03 Release P1 using the x64.GUI_Upgrade.xz method with no problems noted. I have a question about the new CIFS AIO default. The Readme says "tune the AIO size from '1' to 4096". I assume this applies to the Minimum AIO size. What is suggested for the Maximum AIO size setting?

I don't know if there is a suggested maximum size but to be honest, if you're not having any problems with file transfers (you're happy with the speeds) you might consider leaving it at the default. Some hardware was having significant transfer rate issues and this option changed some in an effort to assist this rare situations.

 

[ProtoSD]

Garrett,

I gotta say, I'm not sure if it was you or someone else, but I LOVE the port counter that was added to freenas-common / common showing which port is being built out of X number of ports! Very helpful!

I upgraded my main/physical NAS to 8.03-p1 from 8.02 today with the GUI upgrade and everything went smoothly. Thanks again to everyone involved for the nice job!

 

[Daisuke]

Just upgraded from 8.02 x64 release to 8.03 Release P1 using the x64.GUI_Upgrade.xz method with no problems noted. I have a question about the new CIFS AIO default. The Readme says "tune the AIO size from '1' to 4096". I assume this applies to the Minimum AIO size. What is suggested for the Maximum AIO size setting?

Disable AIO and set to 4096 the Min and Max AIO size. That's what I did, based on the 8.0.3-p1 README info. Are there any other settings that changed from previous default 8.0.x values?

 

[gcooper]

Garrett,

I gotta say, I'm not sure if it was you or someone else, but I LOVE the port counter that was added to freenas-common / common showing which port is being built out of X number of ports! Very helpful!

Yeah.. I thought it helped. nanobsd needed a little bit of love for that ;).

I upgraded my main/physical NAS to 8.03-p1 from 8.02 today with the GUI upgrade and everything went smoothly. Thanks again to everyone involved for the nice job!

Thanks!

 

[gcooper]

Just upgraded from 8.02 x64 release to 8.03 Release P1 using the x64.GUI_Upgrade.xz method with no problems noted. I have a question about the new CIFS AIO default. The Readme says "tune the AIO size from '1' to 4096". I assume this applies to the Minimum AIO size. What is suggested for the Maximum AIO size setting?

Stick to multiples of 4096 (this is _SC_PAGE_SIZE -- multiples of 4kB are good to avoid memory page fragmentation which will improve how things work with Samba when it allocates memory chunks -- this is probably a microoptimization but meh..) and adjust to taste is the best I can say. Without having enough hardware and switch configurations to test with, one size fits all recommendations are hard to prescribe, and FWIW unless you're running hardware that can get near line speeds you shouldn't see a big difference; many home users won't see a difference because pulling music/videos/etc over the network is pretty low bandwidth in the grand scheme of things as far as contemporary hardware goes -- wireless non-withstanding >_>.

 

[Simon00]

I just upgraded to 8.0.3 p1 and noticed write speeds are reduced on a system w/ only 2GB of RAM. Before I got a continuous 35MB/s read/write speeds, but now, writes happen in 2-3 second intervals at 35MB/s and drops to half that speed half the time. I suspect the AIO change is the cause... I guess I need a memory upgrade from now on or change the default.... even though I use it only to backup files w/ only a single computer connected to it.

Was AIO disabled by default in the previous 8.0.3 version?

 

[gcooper]

I just upgraded to 8.0.3 p1 and noticed write speeds are reduced on a system w/ only 2GB of RAM. Before I got a continuous 35MB/s read/write speeds, but now, writes happen in 2-3 second intervals at 35MB/s and drops to half that speed half the time. I suspect the AIO change is the cause... I guess I need a memory upgrade from now on or change the default.... even though I use it only to backup files w/ only a single computer connected to it.

Was AIO disabled by default in the previous 8.0.3 version?

It's in the release notes for 8.0.3 up on SourceForge (unfortunately our other documentation and some other bits are lagging, but I'll send out the note to everyone to get that rectified) :).

 

[Simon00]

I guest for compatibility/reliability reasons Asynchronous IO is enabled by default now....

Is Synchronous IO used if AIO is disabled?

From my understanding Synchronous communications is virtually always more complex but also have much higher throughput than Asynchronous.

With previous 8.0.3 I was getting slightly faster writes than read, but now it's much slower writes w/ AIO enabled.