On routers with MPC-3D- <x> FPCs, it is possible to see <cede> IDMEM Parity error messages.

On routers with MPC-3D- <x> FPCs it is possible to see IDMEM Parity error messages; similar to the following:

Nov 30 17:15:21  MX960-LAB fpc3 LU 2 RD_NACK 2 AP[0x04] TOE Write 0x002913a0
Nov 30 17:15:21  MX960-LAB fpc3 LU 2 IDMEM Parity error in Bank 3, Count 10, IDMEM Bank 3 Offset 0x00014899 IDMEM[0x00052274]
<repeated>

• Highlight if these errors may or may not be hardware related.
• Demonstrate a procedure to determine if they are or not truly hardware related.
• Provide an optional procedure to correct these errors; if proven to be cosmetic and not hardware related.
• To highlight the Problem Report, in which the erroneous parity errors have been corrected.

The JUNOS software defect, which causes these false IDMEM parity errors, has been corrected in PR564760 . One solution is to upgrade the router to a Junos version, which has the changes mentioned in PR564760

Note : If you are running Junos with the changes mentioned in PR564760, the following procudure does not have to be performed.

As an alternative method to correcting the memory error, perform the following procedure:

Messages, such as the following, will be repeated as long as SW encounters the error. These errors are likely due to soft errors, which often occur within uninitialized memory locations. These errors are generally a cosmetic issue.

Nov 30 17:15:21 MX960-LAB fpc3 LU 2 RD_NACK 2 AP[0x04] TOE Write 0x002913a0
Nov 30 17:15:21 MX960-LAB fpc3 LU 2 IDMEM Parity error in Bank 3, Count 10, IDMEM Bank 3 Offset 0x00014899 IDMEM[0x00052274]

 

1. Start a local UNIX shell via the CLI:

    

   admin@bx01> start shell
root@bx01%

2. Start a vty to the associated MPC shell prompt via the router shell:

   root@bx01% vty fpc3

NPC platform (1067Mhz MPC 8548 processor, 1024MB memory, 512KB flash)

NPC3(bx01 vty)#

3. The IDMEM errors will be displayed in the VTY. Disabling the syslog to this TTY will not affect the syslog on the RE; but it will suppress this TTY, making the command entry easier.
   
   Note : These commands must be entered with no typos/mistakes. Verify each command before typing.

   NPC3(bx01 vty)# set syslog tty disable
Logging has been disabled on this tty

4. All memory allocations are registered. Verify if the failing location is not currently in use by the software.

   NPC3(bx01 vty)# show jnh 2 idmem 0x00052274
IDMEM at offset 0x00052274 No allocation ID for memory

   In this example, the memory location is not being used by software; so, this is a cosmetic error message.
   
   If this command shows that the memory is being used by software, the issue is not cosmetic and may indicate a hardware issue. In this case, contact JTAC for further assistance, mention this article, and provide the session log of the above steps. Additionally, do not continue with the remaining steps in this procedure.
   
   If the error is shown to be cosmetic, as identified above, a write to the failing location will safely repair the failing location. Although, in most cases, this command can be safely executed, an error in the typed command may have indeterminate results. So, It is recommended to continue, beyond this point, during a maintenance window. The following steps are optional and necessary only to eliminate the erroneous parity error messages.

• If the memory is not being used by software (as shown by the output of the previous command), the error can be cleared by writing valid data to the location; as shown below:

  NPC3(bx01 vty)# set luchip 2 mem imem 0x00052274 0 1

• Read the memory location to check if the correct value is in this location:

  NPC3(bx01 vty)# show luchip 2 mem imem 0x00052274 1
idmem offset 0x52274 -> 0x52274
0x0000000000000000

• Exit the VTY and shell to return to the CLI:

  NPC3(bx01 vty)# exit
root@bx01%

root@bx01% exit
admin@bx01