OS6 now works on x86 Legacy WARNING: NO NTGR SUPPORT!

walshlink wrote:
Jumbo frames can cause problems and should not be implemented unless there is a specific reason to do so.

Would you please elaborate on that statement?

I've been utilizing Jumbo Frames on two of my ReadyNAS Pro 6's since day one, and can't say I've noticed any significant issues. . .

I'm not sure I'd put it quite the same way as walshlink.

But jumbo frames can certainly cause problems - you need to ensure that all your equipment works with them, and that you pick an appropriate and common MTU. When that is taken care of, it is also quite possible that some your equipment will actually run faster with normal ethernet.

If everything is working, and performance is better with Jumbo frames, then it makes sense to leave them on.

But if you are seeing issues, or if the performance in fact isn't better, then the simplest answer is to turn them off.

StephenB wrote:
If everything is working, and performance is better with Jumbo frames, then it makes sense to leave them on. But if you are seeing issues, or if the performance in fact isn't better, then the simplest answer is to turn them off.

With THAT, I concur!
Thank you for your input. . .

ATCIS, do you have jumbo frames enabled on both your PC and switch? Are your PC, switch, and RNP all using the exact same MTU size?

If you are using your jumbo frame-enabled PC to connect to both your storage device and the Internet, you Internet speeds will suffer as your ISP will fragment the packets down to 1500.

walshlink wrote:
...If you are using your jumbo frame-enabled PC to connect to both your storage device and the Internet, you Internet speeds will suffer as your ISP will fragment the packets down to 1500.

I agree that this can happen. Though MSS negotiation is supposed to prevent it on the uplink for TCP, and websites, etc on the internet shouldn't be sending jumbo frames on the downlink no matter what. In any event, I agree it is a good idea to test internet speeds after configuring JF, as well as speeds to WiFi or fast ethernet devices on your home network. If you use a VPN, then make sure you check that too.

Personally I don't use JF. My pro-6 performs well enough w/o them, and I'd rather use my network than play with its settings. Though JF certainly can be problematic, there are users who find they work fine on their networks.

Where it gets tricky is when you get folks who don't get the limitations, and refuse to try turning them off when they report performance and connection problems.

Bonjour Poussin,

Apres une reinstallation du NAS, j'ai installe directement ton addon CUPS. J'ai note que je n'avais pas access a l'administration cups IP:631 meme apres reboot. En fait il a fallut executer la commande : cupsctl --remote-admin

Peut-etre que tu peux integrer cette commande dans l'addon?

Merci de tes contributions a la communaute.

walshlink wrote:
ATCIS, do you have jumbo frames enabled on both your PC and switch? Are your PC, switch, and RNP all using the exact same MTU size?

I have Jumbo Frames enabled on all devices that support Jumbo Frames (PC's, Switches, NAS Devices, etc.). I have the MTU set at 9000 bytes throughout. . .

walshlink wrote:
If you are using your jumbo frame-enabled PC to connect to both your storage device and the Internet, you Internet speeds will suffer as your ISP will fragment the packets down to 1500.

I can't say that I've noticed any degradation in Internet throughput. I consistently get 36 Mbps down and 24 Mbps up. :woot:

I start experimenting with JF while my ultra2 was under 4.2.x and I didn't notice any change in performance (nearly no difference at all, just 2MB faster or slower depending on read/write or huge/big files. I didn't kept the precise results of my copy tests (I made 3 times each test in order to obtain reliable numbers). But my conclusion at that time was that the differences were too small compare to the variability of the results and/or not worth the risk to have issues with JumboFrames enabled.
Then I switch to OS6 and I notice speed improvements. I don't know if it's due to a better use of the 2GB DDR for caching, or if it was due to OS6/BTRFS file system. But it has improved for sure.
I was very fond of JF.. trying to improve transfer speed for big/huge files but after reading again and again, making my own tests and noticing some strange behavior etc.. I realized that very often it's not worth the risk to obtain issues or hiccups in this 'quest'.
So I didn't bother to do the sped comparison again. One day maybe ... if I have time to loose :D

PS : For those who don't speak french, here is the translation of papagaiou's post :
"Hi Poussin,
Once I reinstalled my NAS, I directly installed your add-on CUPS. I noticed that, even after a reboot, I didn't have access to CUPS's administration IP:631. I had to execute the following command : cupsctl --remote-admin
Maybe you could add this command inside the add-on ?
Thanks for your contributions to this community."

arnomc wrote:
Then I switch to OS6 and I notice speed improvements. I don't know if it's due to a better use of the 2GB DDR for caching, or if it was due to OS6/BTRFS file system. But it has improved for sure.

For what it's worth, my average speed improvements with Jumbo Frames enabled (totally unscientific testing methods) vary between 15 and 25%. One thing I have noticed that makes a significant difference with regard to whether Jumbo Frame yield any benefit whatsoever is how it is implemented in the SWITCH firmware. I've done some testing with older 3Com gigabit switches and newer NETGEAR switches. The older 3Com switches apparently did not have the correct Jumbo Frame implementation, as transfer speeds were either the same or worse (slower) than with "normal" frames enabled.

If I had to guess, folks that notice a negative impact on file transfer speeds when Jumbo Frames are enabled, are not using a compatible (or current) Jumbo Frame implementation in their switch/es firmware.

Just my 2 cents. . .

arnomc wrote:

I start experimenting with JF while my ultra2 was under 4.2.x and I didn't notice any change in performance (nearly no difference at all, just 2MB faster or slower depending on read/write or huge/big files. I didn't kept the precise results of my copy tests (I made 3 times each test in order to obtain reliable numbers). But my conclusion at that time was that the differences were too small compare to the variability of the results and/or not worth the risk to have issues with JumboFrames enabled. Then I switch to OS6 and I notice speed improvements. I don't know if it's due to a better use of the 2GB DDR for caching, or if it was due to OS6/BTRFS file system. But it has improved for sure. I was very fond of JF.. trying to improve transfer speed for big/huge files but after reading again and again, making my own tests and noticing some strange behavior etc.. I realized that very often it's not worth the risk to obtain issues or hiccups in this 'quest'. So I didn't bother to do the sped comparison again. One day maybe ... if I have time to loose :D PS : For those who don't speak french, here is the translation of papagaiou's post : "Hi Poussin, Once I reinstalled my NAS, I directly installed your add-on CUPS. I noticed that, even after a reboot, I didn't have access to CUPS's administration IP:631. I had to execute the following command : cupsctl --remote-admin Maybe you could add this command inside the add-on ? Thanks for your contributions to this community."

no need to do cupsctl --remote-admin on my 516 after a reboot , can you check the administation panel of cups to see if remote admin is ticked ?