Single bit memory errors

[John E. Goetz]

Can someone supply me with the procedure on how to resolve single bit memory errors. I believe I need to be into Single User mode to fix.

[Robert Salter]

John,

I had a server that was getting single bit memory errors and was told by HP that unless they were excessive in short priod of time, that it wasn't a big issue. After a reboot they would go away for a month or so. Reseating the DIMM in question could fix it and if it is just too annoying or is indeed causing trouble then have HP replace it.

I'm attaching a document I got from one of the HP techs, it may shed some light.

Robert

[Robert Salter · Accepted solution]

Okay, I thought I attached a document. Maybe this time it'll work.

[A. Clay Stephenson]

In almost all cases, the "fix" is to replace the memory module. It is extremely rare that reseating the memory will fix this.

[Kevin Wright]

single bit errors are hardware errors, not related to the OS. to fix, replace the DIMM. I wouldn't bother unless you are getting alot of them. > 3 same dimm in 24 hrs.

[Andrew Rutter]

hi john,

you cannot resolve the single bit errors.

Hp uses ECC memory dimms, so that these errors will correct, alot may indicate a failure pending. If you were getting double bit errors then the system may crash or at least hang.

You can view the memory error log with STM or pdcinfo, these will tell you which dimm is at fault. you can also clear the memory PDT by rebooting and stopping at pdc, and ebntering the service menu. most sytems can hold at least 50 errors before its full and needs attention.

It is better to take this as a warning though and get replacement ready and change at planned downtime rather than unplanned and risk the sytem doesnt come back up

Andy

[John_Hancock]

I concur. Single bit memory errors are bad whichever way you look at it. As has been stated the single bit errors are corrected on the way through so that there is no data corruption. However this means that there is a hardware failure in your memory. As such it needs to be replaced. The last thing that you want is for it to fail permenantly at a critical moment.

Arrange for a replacement and schedule an outage.