Recently my syslog has been filling up within a few days, and it's because of a MCE Memory Error (shown in the logs). I just took a snippet but this error happens once a second and quickly balloons the log. It doesn't start immediately on server startup but it doesn't take too long to start seeing it.

 

I've done some research and i thought it was just a bad stick of RAM...but I've now replaced both sticks I had in there and I'm still seeing this same issue. At this point I'm wondering if it's either a bad slot or something else.

 

Any advice?

mce-syslog.txt

Looks like a real RAM problem, either stick or slot or running out-of-spec.

 

Have you done memtest? Try each stick one at a time in each slot.

16 minutes ago, trurl said:

Looks like a real RAM problem, either stick or slot or running out-of-spec.

 

Have you done memtest? Try each stick one at a time in each slot.

 

When I first encountered the problem I did test 1 stick at a time and i thought it was obvious that one stick had issues but then after I replaced that stick, i realized the other stick was also causing the error. I do think I bought all 4 sticks from the same ebay merchant so there is a possibility that he's just selling bad sticks, but their rep doesn't indicate this is likely.

 

I don't think I've done memtest so I will try that.

Are you running the RAM at the CPU rated spec or the XMP profile?

You're running ECC memory.  The memtest that comes with the OS will not display any errors because they're being corrected.  Does your system event log in the BIOS tell you anything more.

 

Set up a new bootable stick via https://www.memtest86.com/ and run it (it's supposed to find ECC errors)

18 hours ago, JonathanM said:

Are you running the RAM at the CPU rated spec or the XMP profile?

 

I haven't modified any RAM settings so whatever the CPU set is what it's running at.

 

I'll try bootable memtest this weekend and see what the results are.