Some years ago, I got a great deal on 3 surplus rx2620 servers, previously used for ~7 years until July 2012 to run HP-UX (according to the SEL).

I've been cannibalizing them for parts to make one really good server, to run OpenVMS (what else would I have bought an Itanium for?), and just recently it started suddenly freezing. It had been in storage at my mom's house (indoors) for several years, and I couldn't get any useful info from the logs as to the cause of the freezes.

I did finally get one alert, priority 7 with a * next to it, but all I can tell from the HP IPMI decoder page is that it was a system firmware error of type HP-specific.

# Location|Alert| Encoded Field | Data Field | Keyword / Timestamp
-------------------------------------------------------------------------------
699 SFW *7 0xC16293B734020370 003FA17000130300 Type-02 137001 1273857 29 May 2022 18:11:00

The decoded version, from analyze/error/elv:

SEL type 2 Subpacket
SEL Record ID 0x0370
TIMESTAMP (Seconds since 1-Jan-1970) 0x6293B734
GENerator ID 0x00C1
EvM Rev 0x03 IPMI V1.0 format
Sensor Type 0x13
Sensor Number 0x00
Event Dir/Type 0x70
Event Type 0x30
Event Direction 0x1
Event Data 1 0xA1
Event Data 2 0x3F
Event Data 3 0x00

Sometimes the offline memory diagnostic would lock up, but no useful messages from any of the diagnostics, including cpudiag. No messages at all from the BMC when turning it on sometimes and no patterns of LEDs animating on the front (very bad signs!). Finally, it did turn on but it said the second CPU had been deconfigured. So that was the root cause: a faulty CPU #1 (or something on the board connected to it, anyway).

I'm a little disappointed that the BMC and diagnostics didn't point me in the right direction, but finally after seeing the second CPU disappear, I decided it was time to put my drives, 12GB RAM, and the quietest fans into another rx2620, and now everything is happy once more.

My only annoyance now is that the new chassis has a very old version of the rx2620 firmware (3.10), which hasn't caused me any problems other than a few missing features (no remote IPMI query over LAN, probably a security risk anyway). Unfortunately, HPE in their infinite wisdom have put the rx2620 firmware ISO and EFI files behind a paywall (support contract), which really annoys me because I downloaded them for free and now don't have the file.

If someone can help me with the rx2620 firmware update, please let me know by email. In the mean time, I'll use the version that it came with and hope there are no serious bugs in the ACPI or platform firmware that affect VMS..

I should have updated all three of the servers at once when I had the chance, but I guess I wasn't thinking clearly in 2013 and didn't expect HPE to split off and start charging money for firmware.

Hopefully this info is useful in the future to someone collecting rx2620 systems as cheap hobbyist machines to run VMS. If your rx2620 is freezing up for no reason, and there's no nearly no useful info from the BMC, it could ultimately be a bad CPU (or perhaps motherboard). Best to buy more than one if you can so you can swap the entire machine out.

On Monday, May 30, 2022 at 5:26:39 PM UTC-7, Jake Hamby wrote:
>
> If someone can help me with the rx2620 firmware update, please let me know by email. In the mean time, I'll use the version that it came with and hope there are no serious bugs in the ACPI or platform firmware that affect VMS.

Never mind about the firmware. I figured out a solution involving the Windows Server 2008 R2 trial CD and the Itanium Support Package drivers and firmware updater from Windows. Fortunately I had a spare hard drive to install Windows just for that purpose.