A few weeks ago, the front LED panel on my PE2900-III turned amber. When I logged into it via DRAC5, the log showed several entries saying the DIMM4 had a lot of errors that exceeded some threshold. I assume this meant DIMM4 might be going bad but ECC was handling the errors so far. I moved DIMM4 to DIMM1 just to confirm the problem followed the FB-DIMM and not the slot or motherboard. Sure enough, after a few days the blue light turned amber and this time DIMM1 was reported as having too many errors. As a quick remedy, I had on hand another FB-DIMM of the same RAM size (8GB), but different brand and different ranks (the original was and Elpida 4Rx4 8GB FB-DIMM, the spare I had on hand is Crucial 2Rx4 8GB FB-DIMM), so I installed the Crucial FB-DIMM. The server booted up fine, and I haven't seen an amber light for almost two weeks.
This week, I'm setting up monitoring on this server, and noticed that OMSA is still reporting DIMM1 in CRITICAL state: # omreport chassis memory Memory Information Health : Critical Memory Redundancy Fail Over State : Inactive Redundancy Configuration : Disabled Attributes of Memory Array(s) Attributes of Memory Array(s) Location : System Board or Motherboard Use : System Memory Installed Capacity : 65536 MB Maximum Capacity : 65280 MB Slots Available : 12 Slots Used : 8 Error Correction : Multibit ECC Total of Memory Array(s) Total Installed Capacity : 65536 MB Total Installed Capacity Available to the OS : 63255 MB Total Maximum Capacity : 65280 MB Details of Memory Array 1 Index : 0 Status : Critical Connector Name : DIMM1 Type : DDR2 FB-DIMM - Synchronous Size : 8192 MB Is there a state I have to reset? Or, is it reporting this because the FB-DIMM is of different rank and/or brand? (same speed) The server has been running fine, no more amber lights, but this is concerning and certainly will setup alarms when the monitoring goes into effect. Just wondering if there is something I need to do to "clear" the "CRITICAL" state, or if I simply need to get an FB-DIMM of the same 4Rx4 rank and brand? Thanks for any suggestions, Bond _______________________________________________ Linux-PowerEdge mailing list Linux-PowerEdge@dell.com https://lists.us.dell.com/mailman/listinfo/linux-poweredge