Intel Atom C2xxx LPC failures

[garyd9]

Can you please share how you're pushing back?  I have an approved RMA for repair… but the turn around time is going to be significant for me..  (Just the 2 way shipping time will be 3 weeks total.)

[doktornotor]

@garyd9:

Can you please share how you're pushing back?

[dopey]

@garyd9: this was my post with a cut and paste of my exact email response when they rejected my cross-ship :)

@dopey:

the RMA person rejected the cross-shipping.  I ordered my motherboard august 2015.

I responded with this:

This was the response i received from technical support regarding my inquiry:
Thanks for contacting us regarding the Atom C2000.

You may submit a RMA at the following link to have the concerned issues addressed.  You may request the board to be cross-shipped during the RMA submission process.

https://www.supermicro.com/support/rma/

They told me I may request cross-ship.

I was under the impression that the intel C2000 flaw couldn't be repaired and needed to be replaced.  My motherboard is still functioning but it's a critical piece of equipment for me (my router) and the downtime would be very inconvenient.

And an hour later I got my cross-ship RMA form :)

[garyd9]

I begged.  It worked.  So, I'll get the replacement board in a few days instead of having to wait a month for roundtrip shipping and repair.

Thank you, SuperMicro, for preemptively replacing a board that MIGHT break in the next 3 years without forcing me to wait until it's broken before addressing it.

The irony is that I've already relocated pfsense to a VM in preparation of a long shipping/repair cycle.  Well, I'll be glad to get the critical router back to dedicated hardware.  It's also good to know that I CAN (if needed) relocate router functions to hyper-v.  (Though it's not quite as easy as I expected due to different interface naming.)

[Derelict]

'll be glad to get the critical router back to dedicated hardware.

If it is critical you should have at least two.

[garyd9]

@Derelict:

If it is critical you should have at least two.

True…  Of course, "critical" in this particular instance is because when the router goes down, my wife/kids whine.  So, probably "critical" might be too strong of a term.

At least the experience has taught me how to get the router working in a VM in case I ever have issues again.  I guess what I need to learn is how to set up a VM so that it can more easily take over the router functions without my having to manually edit config.xml files (to change interface names from igbX to hnX) and then manually restore them.  (I'd still have to manually move the WAN ethernet cable... I think.)

[dennypage]

@garyd9:

True…  Of course, "critical" in this particular instance is because when the router goes down, my wife/kids whine.  So, probably "critical" might be too strong of a term.

When my wife is upset, it's critical to me. That's why I have a spare SG-4860.

:)

[whosmatt]

@garyd9:

At least the experience has taught me how to get the router working in a VM in case I ever have issues again.  I guess what I need to learn is how to set up a VM so that it can more easily take over the router functions without my having to manually edit config.xml files (to change interface names from igbX to hnX) and then manually restore them.  (I'd still have to manually move the WAN ethernet cable… I think.)

CARP and the redundancy features in pfSense work wonderfully for this… provided you have at least three static IP addresses on your WAN... probably you don't on a standard residential connection.  Otherwise, a simple nightly config backup and something like sed will easily provide you with a ready-to-go config file for your backup router (if the hardware is different).  Otherwise just the backup is fine.

[gcu_greyarea]

@Derelict:

'll be glad to get the critical router back to dedicated hardware.

If it is critical you should have at least two.

That's perfect valid point for an enterprise environment.

The SG-2440 is advertised and targeted at:

https://store.pfsense.org/SG-2440/

Small Businesses
Small to Medium Sized Business Networks
Small to Medium Sized Branch Office
Managed Service Provider / Managed Security Service Provider (MSP/MSSP) On Premise Appliance
Teleworkers needing an "Always-Up" network or VPN connections

I fail to see why a small branch office should have a HA Router Setup.

An then there are claims about:
"No moving parts to wear out. This system is designed for a long deployment lifetime."

That should read: "This system is designed for a long deployment lifetime with increased likely hood of failure after 18 months and almost certain failure after 3 years"

Would anyone here purchase a second car - just because the car manufacturer refuses to fix a faulty engine component???

An HA setup is system architecture.

Not fixing a known issue is neglecting customers.

If driving is critical to you - just buy two cars?

[dopey]

Yeah. critical for those of us geeks that have pfsense at home is different than critical for the enterprise :)

My backup plan for my critical piece of equipment is to pull out my old dd-wrt router that's still configured to properly work with about 2/3rds of my stuff, and stick it in place until I can fix my pfsense router :)

Sure, at that point, I'm limited to 200-300mbit/s speeds but I can live with that and I'm sure my wife and kids would never notice :)

As long as they can get to their netflix and local plex server, and I can vpn to work, we're all happy.

[VAMike]

@gcu_greyarea:

I fail to see why a small branch office should have a HA Router Setup.

Because it's an incredibly cheap way to ensure that an office full of people doesn't go idle because there's a network problem. I'd also recommend a backup internet connection if it's at all possible. I guess if salaries are really low in your area then the extra couple hundred bucks isn't worth it, but in most places nowadays a day without internet is going to cause a lot more than a couple hundred bucks worth of loss.

You seem to think that the C2xxx errata is the only thing that can possibly go wrong with a firewall, but most people would be sad if their office went offline for some period of time because (to use an example I've seen personally) their $5 wall wart failed. An HA configuration prevents multiple failure modes.

[gcu_greyarea]

Of course, there are many things that possibly can and will go wrong.
That's why you'd address the obvious and known issues first.

Looks like the C3000 has been announced…. which will most likely replace my C2000 kit. I'm sure the C3000 will have its own quirks, but hopefully intel learned from the clock issue.

I wouldn't want a "reworked" C2000 board anyway (if the soldering is done by a human).
Preference would be the intel fix - or - an entirely new CPU (i.e. C3000)

I decided that next time I will "Roll my Own" Hardware, so that I can blame myself if anything goes wrong. I still like the fanless ADI/Netgate kit, but I lost confidence in the company.

The advantage of the Netgate appliances is that they have no moving parts, don't require assembly and can be deployed quickly. But they are certainly not really that cheap, considering you can get more powerful hardware for the same money.
And despite the generous warranty extension I do not like the fact that Netgate won't proactively replace the systems for affected customers.
Let alone - continues to sell faulty units.

[Derelict]

Wasn't necessarily talking about HA but, since the term critical was used in the context of the unlikely event of an LPC component issue, there should be at least a spare on the shelf. If it's a critical router.

[chpalmer]

@Derelict:

since the term critical was used in the context of the unlikely event of an LPC component issue, there should be at least a spare on the shelf. If it's a critical router.

That is even true if your that afraid of your wife..  ;D

I usually just offer taking mine out to dinner and all is good!    ;)

[garyd9]

For those with supermicro boards who were approved for, and submitted the cross-ship agreement, have any of you had any news from supermicro since returning the cross-ship agreement?

I got the impression from that document (in particular, the part stating "*The replacement will be shipped by FedEx/UPS Standard Overnight for domestic destinations…") that they try to get those cross shipments out and delivered quickly.  If that's the case, at least some of you would have received a replacement board by now...

(I sent my form back yesterday, but it's been over 24 hours and I haven't seen anything back from them.  Not a tracking number, or anything else.)  I'm starting to wonder if they've actually shipped out ANY replacement boards yet (that have the issue resolved.)

That leads me to wonder if the "platform level change" that Intel claims works around the problem requires manufacturing a new revision of motherboards, or if it's something they can do as a "repair."  If the former, I'm wondering if Supermicro is just going to wait for a new stepping from Intel instead of bringing up new tooling/etc to produce new boards.

THAT, in turn, makes me wonder if the supermicro boards that pfsense got as "advance replacements" are actually boards with the problem resolved, or if they are just more with  the same issue, but held in reserve to quickly service failed devices.  (I don't think that tidbit of info was mentioned.)  If the latter, it would certainly explain why pfsense hasn't commented on if "current" stock has the issue already resolved or not.  The answer MIGHT be that current (and even current replacement) stock doesn't resolve the issue, because their vendor hasn't sent them anything yet with the issue resolved.

(Of course, all this is pure speculation and guessing.)

Edit: 45 minutes after posting the above, I got a tracking number from supermicro... which kind off negates this entire post.

[dopey]

@garyd9:

For those with supermicro boards who were approved for, and submitted the cross-ship agreement, have any of you had any news from supermicro since returning the cross-ship agreement?

…

That leads me to wonder if the "platform level change" that Intel claims works around the problem requires manufacturing a new revision of motherboards, or if it's something they can do as a "repair."  If the former, I'm wondering if Supermicro is just going to wait for a new stepping from Intel instead of bringing up new tooling/etc to produce new boards.

...

I received my replacement on Friday and it's already in place.  The board had a Tested 2/21/17 sticker on it from QA.
I honestly have no idea if it has the platform fix or not, but supposedly the platform fix can be retro-fitted.  This is what servethehome has listed for supermicro:

Supermicro: RMA for platform-level workaround available for concerned customers. We also did confirm that Supermicro has implemented the platform level workaround in products shipped from January 2017 onwards.

I suppose at this point I'll have to wait and see :)

FYI - my backup plan?  ordering a 4 port intel nic to stick into my virtual host and setup a backup pfsense instance there.  I'm also thinking of picking up one of those $50 edgerouter x devices to play with :)

[garyd9]

@dopey:

I received my replacement on Friday and it's already in place.  The board had a Tested 2/21/17 sticker on it from QA.
I honestly have no idea if it has the platform fix or not, but supposedly the platform fix can be retro-fitted.  This is what servethehome has listed for supermicro:

Did you notice if it's a REV 1 board, or if they bumped the REV to 02?  Is there any sign of jumper wires on the board?

[dopey]

@garyd9:

Did you notice if it's a REV 1 board, or if they bumped the REV to 02?  Is there any sign of jumper wires on the board?

I found nothing different other than the QA sticker.

I sent an email asking about any confirmation of the fix

Thanks for processing this.  I received the RMA on friday.  There doesn't appear to be any distinguishing marking (rev bump stuff like that) to note whether or not the board has the platform level workaround implemented for the atom cpu flaw.  Is there anyway to get some kind of confirmation that it actually has that workaround implemented?

And this was the response

Hello
The replacement has the issue fixed.

Guess I just have to trust them :)

[baggar11]

Does the new replacement board have a different stepping if you check via command line?

@dopey:

@garyd9:

Did you notice if it's a REV 1 board, or if they bumped the REV to 02?  Is there any sign of jumper wires on the board?

I found nothing different other than the QA sticker.

[garyd9]

@baggar11:

Does the new replacement board have a different stepping if you check via command line?

I got my replacement last night, and I see NOTHING different whatsoever on the board (other than its an obvious refurb that hasn't been as gently handled as my original.)  The CPU stepping is also identical: Origin="GenuineIntel" Id=0x406d8 Family=0x6 Model=0x4d Stepping=8

On my replacement (which was a cross-ship), I've had a few problems already.  I've had to clear CMOS a couple times to get it booting, and then it crashed (kernel crash) in the middle of booting pfsense, which then resulted in a corrupt filesystem (and we all know how poorly pfsense 2.3.x deals with that.)

All of these issues COULD be related to the CMOS being whacked out.

Since then, I pulled the CMOS battery, erased CMOS again (several times), reconfigured BIOS, completely reinstalled pfsense and restored a backup configuration.  So far, it doesn't seem to be doing anything bad… but it hasn't even been 24 hours since I got it working properly.