Memory Error Corrected threshold exceeded

[gugoni jr]

Hy.
I have a blade BL20p G4 with S.O. "Red Hat Linux Enterprise Server 5.1 release" that generated message from memory.

In the logs of IML ILO appears the following error:
"Memory Error Corrected threshold exceeded (Memory System, Memory Module 7)."
That would be a mistake of memory or information.

Another thing, how do I delete the attention led to the blade?

Thanks.

[JKytsi · Accepted solution]

Clear the IML log.

[Blazhev_1]

Hi,

this is there to tell you that soon the memory can fail and you will have to plan downtime to replace it...

Pac

[gugoni jr]

Jarkko K.
Even clearing the log IML not deleted the led.

[gugoni jr]

Pac, thank´s for information.
I wonder if to erase the attention led on the blade, is only reboot?

[Blazhev_1]

Hi,

yes , reboot will maybe reset the LED, but soon it will turn on again.

When you write 8 bits to the DIMM and 1 is changed (0 instead of 1), the parity detects this and corrects 1 bit(ECC). There is a threshold for the allowed errors. For example if for 3 days you have 20 errors on the same module, something is not ok and the System ROM tells you to replace the DIMM ASAP, before it becomes critical. When you reboot the server , you will reset the threshold counter, but this will not prevent the DIMM to fail or to turn the LED on if the threshold exceeds again.

I don't know another way to turn the LEd off

Pac.

[kovvu]

When I encountered this problem, replacing a bad DIMM make the error message disappear.

[Jerry Schafer]

I have several of these blades in my IT operation. Bottom line: The LED is telling you that there is something wrong with the memory DIMM. You can delete the IML entries but you cannot turn off the LED. The way to turn it off is by fixing the problem. I suggest that you power down the blade and reseat all the memory DIMMs (remove the DIMMS, make sure the sockets are clean, re-insert the DIMMS). This will often take care of the problem. If that does not fix the problem, the DIMM needs to be replaced. In my case, where the blades were still under HP warranty, I reported the problem to HP and they took care of the problem by sending a replacement DIMM.

[gugoni jr]

Resolved.