RaptorCS's redemption: the POWER9 machine works October 10, 2019 on Drew DeVault's blog

This is a follow-up to my earlier article, “RaptorCS POWER9 Blackbird PC: An expensive mistake”. Since I published that article, I’ve been in touch with Raptor and they’ve been much more communicative and helpful. I now have a working machine!

Update Feb. 2024: Seems the improvements I asked for may not have stuck. Buyer beware.

Picture of uname -sm showing “Linux ppcle64”

After I published my article, Raptor reached out and apologised for my experience. They offered a full refund, but I agreed to work on further diagnosis now that we had opened a dialogue1. They identified that my CPU was defective and sent me a replacement, then we found the mainboard to be defective, too, and the whole thing was shipped back and replaced. I installed the new hardware into the datacenter today and it was quite pleasant to get up and running. Raptor assures me that my nightmares with the old board are atypical, and if the new board is representative of the usual user experience, I would have to agree. The installation was completely painless.2

However, I refuse to give any company credit for waking up their support team only when a scathing article about them frontpages on Hacker News. I told them I wouldn’t publish a positive follow-up unless they also convinced me that the support experience had been fixed for the typical user as well. To this end, Raptor has made a number of substantive changes. To quote their support staff:

After investigation, we are implementing new mechanisms to avoid support issues like the one you experienced. We now have a self-serve RMA generation system which would have significantly reduced your wait time, and are taking measures to ensure that tickets are no longer able to be ignored by front line support staff. We believe we have addressed the known failure modes at this time, and management will be keeping a close eye on the operation of the support system to ensure that new failure modes are handled rapidly.

They’ve tweeted this about their new self-service RMA system as well:

We’ve made it easy to submit RMA requests for defective products on our Web site. Simply go to your account, select the “Submit RMA Request” link, and fill out the form. Your product will be warranty checked and, if valid, you will receive an RMA number and shipping address!

— @RaptorCompSys via Twitter

They’re also working on other improvements to make the end-user experience better, including more content on the wiki, such as a flowchart for dealing with common problems.

Thanks to Raptor for taking the problem seriously, quickly fixing the problems with my board, and for addressing the systemic problems which led to the failure of their support system.

On the subject of the working machine, I am quite impressed with it so far. Installation was a breeze, it compiles the kernel on 32 threads from spinning rust in 4m15s, and I was able to get KVM working without much effort. I have christened it “flandre”3, which I think is fitting. I plan on bringing it up as a build slave for builds.sr.ht in the coming weeks/months, and offering ppc64le builds on Sourcehut in the near future. I have another board which was generously donated by another Raptor customer4, which arrived last week and that I hope to bring up and use for testing Wayland before introducing it to the Sourcehut fleet.

P.S. For those interested in more details of the actual failures:

This machine is so badly broken that it would actually be hilarious if the manufacturer had been more present in the troubleshooting process. I think the best way to sum it up is “FUBAR”. Among problems I encountered were:

In the immortal words of James Mickens, “I HAVE NO TOOLS BECAUSE I HAVE DESTROYED MY TOOLS WITH MY TOOLS.” A staff member at Raptor tells me: “Your box ended up on my desk […] This is easily the most broken board I’ve seen, ever, and that includes prototypes. This will help educate us for a while to come due to the unique nature of some of the faults.”

Not sure what can cause such an impressive cacophony of failures, but it’s so catastrophic that I can easily believe that this is far from typical. The hardware is back in Raptor’s hands now, and I would be interested to hear about their insights after further diagnosis.

  1. They did refund the RAM which was unfulfilled from my original order. ↩︎

  2. They did give me a little heart attack, however, by sending the replacement CPU to me in the same box I had returned the faulty CPU back to them with - a box which I had labelled “BAD CPU”. ↩︎

  3. Sourcehut virtual machines are named after their purpose, but our physical servers are named after Touhou characters. ↩︎

  4. This happened prior to any of the problems with the first machine. ↩︎

Articles from blogs I read Generated by openring

Status update, May 2024

Hi! Sadly, I need to start this status update with bad news: SourceHut has decided to terminate my contract. At this time, I’m still in the process of figuring out what I’ll do next. I’ve marked some SourceHut-specific projects as unmaintained, such as sr.ht-…

via emersion May 21, 2024

Automatic case design for KiCad

I don't generally get along great with CAD software with the exception of KiCad. I guess the UX for designing things is just a lot simpler when you only have 2 dimensions to worry about. After enjoying making a PCB in KiCad the annoying for me is alwa…

via BrixIT Blog May 15, 2024

The floor is lawa!

And now for something completely different… When was the last time you were excited about a simple window with nothing but a single background color? Well, I currently am. Let me tell you about it… This window is notable, because it was created using the ”pu…

via blogfehler! May 8, 2024