FX2s power supplies

Hi,

Could you check the FX2s's CMC firmware and FC630's iDRAC is up to date? You may get error on the if the both firmware are mismatched. Let's get everything on the firmware up to date, and see if it resolves the issue.

Let me know if you have any questions.

Good morning Subnet,

Does the CMC's log have any errors displayed in it, to your knowledge? You should be able to find that from the web UI.

I took a look at the part numbers you provided, there doesn't look to be an issue there to me, either. With the reporting saying the issue is between the chassis and the node in slot 4, I'd say the next step would be to acquire logs. If you're able to get into the iDRAC of the FC630 and export a SupportAssist collection, it may be enlightening. However, I'd expect the more actionable information to come from the chassis logs. To acquire those, you'll want to SSH into the CMC and run the following two commands 'racdump' and then 'dumplogs.' 

If you're on a Windows workstation, I generally recommend using PuTTY and using it's native option to save output to a text file. This should provide us with a better idea of what the problem is from the chassis' perspective and from the node's perspective.

They are all up to date. CMC is at 2.30 and all iDRACs are at 2.75.75.75 (17).

Here are the requested logs zipped up: https://www.dropbox.com/s/ouzku89qcwvypkh/FX2Logs.zip?dl=0

I should also note that the 1100W power supplies are not currently installed - I didn't want to chance erratic behavior. I do have a second 'cold standby' chassis I could put them in and pull the same logs from if that would be of more help.

Unfortunately, there's not much I can do with these two logs, though I did look at them. I can see that there was an issue previously, but can't provide any constructive recommendations. What I'd need to see is a log that had the questionable power supplies interacting with the node, so that we have the best opportunity of recording the root issue.

Ok, I installed the 1100W power supplies in my second chassis and moved an idle compute module over to it and got the same results. Here are the results from that chassis: https://www.dropbox.com/s/g81aatjy5tnma0h/FX2sLogs2.zip?dl=0

Also, here's a SupportAssist bundle from the FC630: https://www.dropbox.com/s/yje5vr00d9yifyq/TSR20200814114912_43M.zip?dl=0

Both PSUs are showing up correctly in both the server and chassis logs. However, I did see this entry:

Are you changing the PSUs out with the chassis live? If so, what I'd recommend (when time permits) is to shut the chassis down, unplug it from power, hold the power button down for 10-30 seconds, then power it back up. I'm not seeing any other issues, so I think there's a fair chance this will clear it up.

No, this was from a cold chassis. Installed the power supplies, connected them to power, powered on the chassis, then powered on the FC630.

I also just tried starting it up with just the NTCWP power supply installed and got the same mismatch error, so it would seem it's not a mismatch between the two power supplies.

Despite this particular chassis being one that hadn't been powered in quite a while before installing the 1100W power supplies, I tried the unplug it and hold the power button trick, and still no good. As soon as the server is powered up, I get the same 'A hardware incompatibility is detected between Server in slot 1 and PSU' message. 

Any other thoughts? Do you have any 'official' part numbers for the power supplies for the FX2s?

The part number you have is a good one, and having different PSUs shouldn't be an issue. This might be a bit of a longshot, but if the BIOS is behind on the blade, you might try updating it. That said, if your CMC and iDRAC were already up to date, I imagine your BIOS is, too. 

This leads me to believe you must have been sent a bad power supply, especially since we see the behavior in at least 2 chassis.

EDIT: For 1100w power supply parts, here are a few samples.

38GYJ, 9TMRF, CMPGM, GDPF3, HT6GX

The blade I have in the second chassis for testing is current at 2.11.0, and the first one that I had powered up in my main chassis that started all this is only one revision behind at 2.10.5.

With regards to compatibility, is it (or should it be) essentially if it's a 12th gen or 13th gen power supply (in other words, if it's listed as compatible with R620/R720 or R630/R730 etc), fits in the slot and is 1100W, 1600W, 2000W or 2400W, Dell considers it compatible with the FX2s? 

The power supplies are considered chassis components, not server components and I've seen no indication that there would be a PSU compatibility issue. I also checked the chassis documentation and looked for any KB articles to show if this might be an issue, but found nothing to indicate it would be a problem.

I would agree that power supplies are, or at least should be, chassis components, and as such, you'd think that if the chassis is happy, everything else will be as well, but I've found something interesting that may explain what's going on.

In looking for info and firmware updates for the power supplies, I found that if I search for the '95HR5' 1600W power supply and look at the firmware updates in the 'Compatible Systems' section, it's all 13th and 14th gen servers. More or less for the 'J5WMG' 2000W power supply, except that shows only 14th gen servers. For the NTCWP, GDPF3 and the 38GYJ, GDPF3 & HT6GX provided earlier, those all come up with 12th gen servers. The CMPGM lists 13th and 14th gen servers and the 9TMRF is 13th gen servers.

So even though, in theory, the servers shouldn't care about the power supplies as long as the chassis is happy, it would seem that they do in fact care very much....

Good find. I obviously didn't see this same bit of information.