Fake server cards

[jgreco]

jgreco submitted a new resource:

Fake server cards - They definitely exist

Some time ago, forum frequenter @artlessknave kindly sent me a pair of LSI HBA cards that had failed to work out in a FreeNAS build. Having suspected that these were fakes, I asked for them to be shipped here for comparison. However, COVID intervened and I never got around to writing a resource that covers this topic.

There is the original discussion thread at...

Read more about this resource...

 

[Ericloewe]

The hole placement of the presumed knockoff isn't as precise, and the holes are just a bit smaller.

It sorta looks like the slot cover had its drawing laid out by hand, resulting in the awkward pattern, with missed holes and crazy misalignments.

And the silk screening, it just doesn't look ... sharp. Look at the LSI cards and the precise hi-def lettering.

Hard to tell in your pictures, but it adds to the picture that someone re-did all the masks for the PCB layers, silkscreen, slot cover, etc. Especially if they used a dot-matrix silkscreen process to save the cost of manufacturing a physical mask.

 

[Ericloewe]

Sidenote: Intel X540-T2s with weird jacks that don't seem to lock with any normal 8P8C jack we could grab around the datacenter, from the El-Cheapo to reasonably fancy ones: Any thoughts, comments or insight?
The story I'm told is that they were acquired separately from the Lenovo servers they went into, for reasons of cost savings. They've been working pretty well (three of them in use), but we were repurposing one of them from a dead server (temporary bodge while we wait for new DACs) and came across this bizarre phenomenon. Neither of the two ports would cause a plug to lock, no matter what we tried.

 

[jgreco]

Yeah, my goal here is to use this as motivation to pull up some nice hi-res pictures of the various legit LSI inventory I have, the two fake cards from @artlessknave , and maybe even venture back a decade or two to cover my nice Intel DAUL cards, which have some tells when compared to legit ones.

I think this is of value to the wider community, as well, but I don't know if I have the time and energy to reach out to AoS, who has clearly done some work along these lines and has other examples. However, this is one of those things where many eyes help to spot sometimes minor differences, and it is in the interests of not only the TrueNAS community, but also the wider ZFS community, the Proxmox community (bit of overlap I suppose), legitimate eBay sellers like AoS who are getting royally screwed because of the bad reputation that fakes bring. That's why I took the unusual step of linking to an eBay'ers YouTube video, as it is more important to screw the counterfeiters than it is to worry about who is profiteering off YouTube or eBay sales or whatever.

 

[artlessknave]

hey, 4 years late is better than never! I don't have legit 9240 cards myself, or I'd offer some pics of them.

 

[Papa]

This made me go back and look at two cards I have - one failed in short order - the other would not take a firmware upgrade - they are both fakes. I haven't a need to build anything other than finish my sandbox just now, but this is ALL good information.

 

[artlessknave]

two cards I have

which 2 cards do they claim to be? and may we have some pics of them, including the model# claimed?

 

[NugentS]

I posted some cards on: https://www.truenas.com/community/threads/fake-or-real.106109/

I THINK 2 are fake and 1 is real. Unfortunately I am unsure which is which, and it may be the other way around

 

[artlessknave]

I THINK 2 are fake and 1 is real.

I think you are correct

 

[jgreco]

A thread in which a fake Intel card caused lots of problems before being replaced:

Transferring file via share results in a corrupted file with different checksum each time

I use a SMB share to access my files through ethernet on my local network. Recently, when I tried to copy some files, they all ended up corrupted. For example, this image is obviously corrupted If I use 7zip to get the SHA256 value of this file, it will return a different result each time...

www.truenas.com

 

[Ericloewe]

Interesting that they printed the Intel logo on that one...

 

[MrGuvernment]

Sucks this happens, when you spend time and money to get the hardware and then waste time, but finding ways to validate something is fake sure is helpful, has anyone claimed refunds from ebay sellers if they do find they got a fake?

I am curious if one of my Intels might be a fake cause it worked fine for several months then suddenly just seemed to stop..

Any word on if Chelsio T580's have any fakes going around? Just got a T580-SO-CR off ebay to be delivered in about 2 weeks..

 

[artlessknave]

Chelsio

this is the only one I remember seeing that looked suspicious to me of any chelsio card.

Dell Chelsio T540-CR 4-Port 10GBe SFP+ PCIe Full Height Network Card PMXRW

 

[xnaron]

Here is my 9305-24i card I bought off a seller in China in Feb 22. Had the card in a machine for about 1 year but hadn't used it as I was still using netapp jbods. Anyhow went to migrate the disks to this card and realized it is not functioning properly. Does this look like a fake?

 

[jgreco]

It's hard (at least for me) to tell, most of the time I look at a legit card for A/B comparisons. Noted on your card: missing LSI logo (or Avago/Broadcom) on the PCB, barcode stickers on the back are fishy looking - especially the "9305-24i". All of the stickers feel rather unbusy whereas they are typically info-dense on legit cards. Missing any declaration of conformity stickers (the weird "FCC" sticker for US class B, CE, etc). Other presumed-legit examples of the 9305-24i seem to have labels on each of the SFF ports. You might want to see if you can find good pictures of other known-sourced-from-Broadcom cards and look for telltales.

 

[xnaron]

Yes. Unfortunately it is really hard to find real images of this card online. I did research it before I bought it and even called broadcom. They didn't have the serial for it on file but could not rule it a fake as there were possibilities in play where they might not have the serial. The card died over the course of a year. It was plugged into an active system but not actually used. I felt the same way about the stickers initially but all the samples I could find at the time looked similar.

 

[jgreco]

The "died over the course of a year" suggests fake but also does not preclude cooling issues that can turn a legitimate LSI card into an EZ-Bake oven that then cooks itself. The LSI cards are little embedded computers in their own right, running a custom firmware that passes messages from the driver on the host server to the embedded CPU, commands such as "read block 1234567 from disk 12". The embedded CPU then talks down the SAS lanes to fulfill the request. When the LSI CPU gets too hot, these commands, or even worse, the data block being acted on, may become corrupted in transit. This can be a disaster for ZFS because ZFS is often working on many drives simultaneously for RAIDZ or mirror operations. You may wish to consider whether you had possibly placed this card in a location with insufficient airflow. These cards are designed primarily for use in rack mount servers, and these usually have a fan wall in the middle of the chassis to guarantee good airflow over all of the cards.