210 - Quick Path Interconnect Link Degradation on DL580 G7

[Sven Vollmann]

Hello,

since two weeks we have a problem with one of our HP Proliant DL 580 G7 server.

When it boots it shows the warning "210 - Quick Path Interconnect (QPI) Link Degradation. A QPI link is operating in a degraded performace state.". We have called our hardware supporter. The technician changed all boards (both I/O-boards and the board with the BIOS/ILO on it), switched the CPUs and RAM-Cages to identify the damaged component, changed the complete front drawer, the Power-Supplies and finally even the fans. He put out all the network/fc cards and starts the server. Nothing. The warning still appears. Now the technician becomes desperate ;-). Does anyone have another good idea to solve our problem ?

Our BIOS version is P65 10/01/2013.

Sincerely yours

Sven

[ScottNolan]

We had the same problem with one of our DL580 G7 servers, first noticed it after adding a lot of physical memory (went from 160GB on four externder cages to 512GB in eight extender cages; then every reboot or POST we'd see this degraded performance error:

"POST Error: 210 - Quick Path Interconnect (QPI) Link Degradation. A QPI link is operating in a degraded performace state."

We did not have support for our HPs - but when I checked the serial numbers against HP's website I discovered we were still under warranty, yay!  Opened a case and HP first recommented running HP-SUM (Service Pack) to upgrade BIOS, iLO Firmware, and any daughter card firmwares...  we did so - and still had the problem.

So then HP sent out hardware support and replaced "6 parts together: System board, CPU Mem Drawer Board and all 4 CPUs together."

That fixed the problem - no more error on POST and server is happy again.  To be fair - throughout we could boot and run the server, in "degraded performace state" whatever that means... but since this is a critical DB server for us - we made sure it has best performace we can get.

Good luck with your own QPI issues.

[303Tek]

Sorry for refloating this old topic. This is just in case a poor soul like me happens to arrive here looking for a solution:

DL580 G7 servers are popular nowdays in home labs as you can score one for cheap on eBay.

My DL580 booted fine always but with the referred "POST Error: 210 - Quick Path Interconnect (QPI) Link Degradation. A QPI link is operating in a degraded performace state.", so I started my investigation.

NOTE: I was unable to find references of "POST Error: 210" in the DL580G7 documentation.

SOLUTION (in my case): Found out that one of the CPU sockets and the CPU itself was contaminated with remains of dry thermal paste.

Cleaned all the 4 CPUs thoroughly with one drop of isopropyl alcohol and a lint free cloth (Please, don't use paper!) . Also removed a bit of the mentioned dry thermal paste of the bottom of one of the sockets: Thin needle, magnification glasses, lots of light and steady hand. After reasembling everything., booted the server  without errors.

Hope somenone find this information useful.

EDIT: Clarfication, rephrasing, typos

[Behemot]

Mine was also dirty as hell all around the processors (really, dumb ppl should not touch electronics and thermal compounds), but that was not it. This message appeared only after I've added some memory modules (still waiting for 64 16GB modules) and the drawer was likely not attached properly. So reaseating the drawer and pushing it a few times (just to be sure) mitigated the problem. Sometimes it is as easy as that