TrueNAS 13.0-U3 is Now Available

[Patrick M. Hausen]

Well and somehow i can't edit my post anymore too...

There are stricter rules applied to the "Announcements" forum. Possibly iX should lock it entirely to moderators and link to a discussion thread in the public one in the future.

 

[Jailer]

I just updated and it says I'm on 13.0-U3.1. Was there a hotfix slipped in there somewhere or did I miss some announcement?

 

[anodos]

I just updated and it says I'm on 13.0-U3.1. Was there a hotfix slipped in there somewhere or did I miss some announcement?

We published a hotfix because significant number of users were hitting an SMB_ASSERT() while performing SMB2 ioctls on alternate data streams (upstream issue that will be fixed in next upstream release).

 

[MortyMars]

Hi

Is the bug below concerning the version 13.0-U2 corrected?
And can I safely proceed with this update knowing that I have a network card 2.5 GigE Realtek?

"Due to a bug with an upstream networking driver causing data corruption issues with iSCSI sharing configurations, 2.5GigE Realtek NICs are unsupported in 13.0-U2 by default. Warning: at a risk of data corruption, especially if the system is used for iSCSI sharing, the offending driver can be manually loaded. See the Known Issues entry for NAS-117663 for more details and the workaround"

Thank you for this confirmation

Youhou
Is there anyone in the forum to answer my question?

 

[Ericloewe]

Speaking for myself, questions about Realtek are no fun at all. They're like wrestling with a pig, you get dirty and the NIC laughs at you in the end. Just don't use Realtek.

 

[MortyMars]

Speaking for myself, questions about Realtek are no fun at all. They're like wrestling with a pig, you get dirty and the NIC laughs at you in the end. Just don't use Realtek.

Hi Ericloewe,

Thank you for your interest in my problem.
Not using Realtek is nice, but when that's what you have on your motherboard how can you do otherwise?
There is no problem with 13.0-U1.1, and the malfunctions appeared with 13.0-U2.
Does TrueNAS really not want to fix this bug in the 13.0-U3?
It seems to me that this is necessary because I am obviously not alone in this case.
Is there a good channel to report this?

 

[Ericloewe]

Not using Realtek is nice, but when that's what you have on your motherboard how can you do otherwise?

Well, the first part is not buying motherboards with Realtek NICs (aside from the dedicated IPMI NIC, where it's fine). The literal answer to your question is "install an Intel NIC".

It seems to me that this is necessary because I am obviously not alone in this case.

"Necessary" is a strong word. Nobody forced you to buy a Realtek NIC - in fact, all advice is not to. In a rainbows and unicorns world, sure, but I'm sure there are many other drivers that would benefit from developer time with much better potential results.

Is there a good channel to report this?

You can make your case in the bug tracker, though that won't do much, since it's a FreeBSD thing.

 

[Davvo]

Hi Ericloewe,

Thank you for your interest in my problem.
Not using Realtek is nice, but when that's what you have on your motherboard how can you do otherwise?
There is no problem with 13.0-U1.1, and the malfunctions appeared with 13.0-U2.
Does TrueNAS really not want to fix this bug in the 13.0-U3?
It seems to me that this is necessary because I am obviously not alone in this case.
Is there a good channel to report this?

The issue does exist in U1.1 as it lies in FreeBSD, it was just addressed in U2 by iX.

Due to a bug with an upstream networking driver causing data corruption issues with iSCSI sharing configurations, 2.5GigE Realtek NICs are unsupported in 13.0-U2 by default. Warning: at a risk of data corruption, especially if the system is used for iSCSI sharing, the offending driver can be manually loaded.

You can manually load the drivers if you don't use iSCSI as it appears to be (relatively) safe.

 

[jgreco]

Not using Realtek is nice, but when that's what you have on your motherboard how can you do otherwise?

TrueNAS doesn't promise to work with every bit of hardware you might choose to buy. Just like buying a car, if you buy a car that isn't rated for towing and doesn't come with a hitch, you're not going to have a good experience trying to tow a large RV. If you want the towing capacity, you can buy a vehicle that is designed for the task. Computer gear works like this too. If you buy a mainboard with a crappy Realtek on it, it is just never going to be very good at server stuff. The Realtek chipset is not designed for server use and is barely suitable for client-grade use such as web browsing in Windows.

 

[Whattteva]

TrueNAS doesn't promise to work with every bit of hardware you might choose to buy. Just like buying a car, if you buy a car that isn't rated for towing and doesn't come with a hitch, you're not going to have a good experience trying to tow a large RV. If you want the towing capacity, you can buy a vehicle that is designed for the task. Computer gear works like this too. If you buy a mainboard with a crappy Realtek on it, it is just never going to be very good at server stuff. The Realtek chipset is not designed for server use and is barely suitable for client-grade use such as web browsing in Windows.

Totally agree with this analogy especially for servers, but I'd probably take exception with the last sentence cause I've been using mostly Realtek NIC's on all my Windows gaming desktops for at least the last... 15-20 years. And I'd say, they've been very solid and this is in a competitive gaming environment where latency is important (unlike web browsing). Of course, I'm talking strictly about the ethernet cards, not WiFi cards.

 

[Ericloewe]

Let's be honest though, the gamer crowd is far closer to the audiophool crowd than either group will admit. See, for instance, the "Killer" 31337 g4m3r nonsense that was en vogue a decade ago and finally trailed off like five years ago.
Realteks mostly work on Windows because the driver holds things together. No such care and attention was given to the Linux or FreeBSD drivers.

 

[jgreco]

And I'd say, they've been very solid and this is in a competitive gaming environment where latency is important (unlike web browsing).

You can take exception all you want, but you've inadvertently just exposed a weakness. The point of server grade ethernet chipsets is that they're optimized for coping with a high number of simultaneous streams, including TCP offload, interrupt coalescing, virtual function support, multiple queues, and doing this with relatively low latency.

Web browsing tends to be a small handful of simultaneous streams, a task which does not necessarily require assistance in the form of dedicated silicon for TCP offload or most of those other "server" features.

Competitive gaming tends to be oriented towards a low latency single stream of packets, which is relatively easy to deal with. You do not need really need hardware assistance for this, which is why the Realteks have been "very solid" in that environment.

Both of these are substantially easier workloads for the ethernet chipset than many server workloads, where server chipsets can be asked to support hundreds or thousands of parallel streams at once.

It used to be that web browsing only consisted of two or three simultaneous TCP streams, but over time we have seen this grow to a somewhat larger number. It's still not hundreds or thousands, though. As the speed of Internet connections increases, and the complexity of things that we run in the background on a PC increases, the limits of the Realtek ethernet chipsets get closer and closer. At some point, you're going to feel that today's Realtek ethernet chipset is not performing well, just as you would probably feel that yesteryear's 10/100 Realteks really sucked if you tried to use them today, even if you only needed less than 100Mbps.

 

[morganL]

Hi Ericloewe,

Thank you for your interest in my problem.
Not using Realtek is nice, but when that's what you have on your motherboard how can you do otherwise?
There is no problem with 13.0-U1.1, and the malfunctions appeared with 13.0-U2.
Does TrueNAS really not want to fix this bug in the 13.0-U3?
It seems to me that this is necessary because I am obviously not alone in this case.
Is there a good channel to report this?

Hi @MortyMars
Just responding for iX.

The problem is known: https://ixsystems.atlassian.net/browse/NAS-117663

The driver for FreeBSD is a realtek problem. If they don't support it well, we can't use it.

The driver was disabled because corrupting iSCSI data is a huge problem that we want to minimize. I assume the community understands this is the best course of action if we want to protect data. It can only be fixed if Realtek fixes their FreeBSD driver. So far, we don't know of a fix.

There are probably two workarounds:

1. if you aren't using iSCSI, we left instructions for loading the Realtek driver - use at your own risk. If you are using iSCSI, please do not do this.
2. TrueNAS SCALE is based in linux. Realtek drivers for Linux are probably better supported. You might want to ask the SCALE forum if they have had any issues with Realtek drivers.

 

[Whattteva]

You can take exception all you want, but you've inadvertently just exposed a weakness. The point of server grade ethernet chipsets is that they're optimized for coping with a high number of simultaneous streams, including TCP offload, interrupt coalescing, virtual function support, multiple queues, and doing this with relatively low latency.

Web browsing tends to be a small handful of simultaneous streams, a task which does not necessarily require assistance in the form of dedicated silicon for TCP offload or most of those other "server" features.

Competitive gaming tends to be oriented towards a low latency single stream of packets, which is relatively easy to deal with. You do not need really need hardware assistance for this, which is why the Realteks have been "very solid" in that environment.

Both of these are substantially easier workloads for the ethernet chipset than many server workloads, where server chipsets can be asked to support hundreds or thousands of parallel streams at once.

It used to be that web browsing only consisted of two or three simultaneous TCP streams, but over time we have seen this grow to a somewhat larger number. It's still not hundreds or thousands, though. As the speed of Internet connections increases, and the complexity of things that we run in the background on a PC increases, the limits of the Realtek ethernet chipsets get closer and closer. At some point, you're going to feel that today's Realtek ethernet chipset is not performing well, just as you would probably feel that yesteryear's 10/100 Realteks really sucked if you tried to use them today, even if you only needed less than 100Mbps.

That's really what my exception was though. I basically agree with everything you say for servers use case except for the last bit about casual web browsing.

In my opinion, Realtek NIC's are perfectly fine for casual Windows desktop users (which is probably like 80-90% of desktop users).

 

[Whattteva]

Let's be honest though, the gamer crowd is far closer to the audiophool crowd than either group will admit. See, for instance, the "Killer" 31337 g4m3r nonsense that was en vogue a decade ago and finally trailed off like five years ago.
Realteks mostly work on Windows because the driver holds things together. No such care and attention was given to the Linux or FreeBSD drivers.

Lol, This is the first time I've heard about audiophool. Had to Google it.

 

[jgreco]

In my opinion, Realtek NIC's are perfectly fine for casual Windows desktop users (which is probably like 80-90% of desktop users).

You're being kind of confusing. I said, above,

The Realtek chipset [...] is barely suitable for client-grade use such as web browsing in Windows

It works fine in this non-stressy environment because it isn't having to do very much. Handling a stream or even several of traffic is not too challenging and can be accomplished in software. When the system starts to get stressed out, packets are lost and traffic slows, but users are used to the Internet being slow and crappy, and tend to place the blame accordingly.

This basically gets back to what one considers "suitability". It seems like you are arguing your Realtek Ford Pinto is suitable because it has four wheels and can do the minimum competent bit of rolling around, steering, and coming to a halt. I feel that this is minimally suitable functionality for a car and I am not going to call it "perfectly fine." My dad would call it "perfectly fine" though. Heh.

 

[jgreco]

audiophool.

I didn't know there was a formal term for this now. We usually referred to them as Monster Cable idjits.

Made me wonder if I went into the wrong business. Is it possible that one could place hot rod flame decals on an ethernet cable and boost its value by 100x?

 

[Ericloewe]

Is it possible that one could place hot rod flame decals on an ethernet cable and boost its value by 100x?

If audiophools will buy gold-plated plastic optical fiber cables for SPDIF over TOSLINK, I'm sure gamers will buy special low-latency Cat. 8 cables to connect their PoS Realtek NIC to their PoS ISP-provided NAT gateway at special prices*.

* Special for the seller, not so much for the buyer.

 

[jgreco]

If audiophools will buy gold-plated plastic optical fiber cables for SPDIF over TOSLINK, I'm sure gamers will buy special low-latency Cat. 8 cables to connect their PoS Realtek NIC to their PoS ISP-provided NAT gateway at special prices*.

* Special for the seller, not so much for the buyer.

Does it have to be gold-plated?


Our local boys for the win.

 

[Ericloewe]

Not with real gold, but they expect a shiny gold-like finish that can blind innocent bystanders by reflecting the RGB Fans' light in weird direction.