Description

This article explains the meaning of the "CPU Error code: 0x280003", "CPU Error code: 0x280002", and "CPU Error code: 0x280001" alarms and also provides the steps to resolve them.
 

TLDR

  1. 0x280001 - CPU uncorrectable memory error
    1. Impact: MPC may reboot
    2. Suggest Action: replace the MPC as soon as possible
  2. 0x280002 - CPU correctable memory error at rate <= 255 events/24 hours. This alarm will be cleared automatically if there were no errors during the last 24 hours.
    1. Impact: None
    2. Suggest Action:
      1. If the alarm is self cleared, no ACTION is needed
      2. If the alarm is active for more than 48 hours, RMA the board
  3. 0x280003 - CPU correctable memory error with rate > 255 events/24 hours - Major error
    1. Impact: None
    2. Suggest Action: Maybe a sign of memory weakness. You can restart the MPC to clear the alarm, If the alarm persists after restart, replace the MPC when convenient

Symptoms

Syslogs and Alarms

The alarm and logs for "CPU Error 0x280003" are given below:

Active Alarm

user@host> show chassis alarms
1 alarms currently active
Alarm time Class Description
2020-06-09 22:01:11 UTC Major FPC 2 Major Errors - CPU Error code: 0x280003

​Logs

Jun  9 22:01:11  alarmd[72819]: Alarm set: FPC color=RED, class=CHASSIS, reason=FPC 2 Major Errors - CPU Error code: 0x280003
Jun  9 22:01:11  alarmd[72819]: Alarm set: FPC color=RED, class=CHASSIS, reason=FPC 2 Major Errors - CPU Error code: 0x280003
Jun  9 22:01:11  fpc2 Cmerror: Draining ASIC error message queue
Jun  9 22:01:11  fpc2 cmerror_process_queue: module = CPU Error

The alarm and logs for CPU Error 0x280002 are given below:

Active Alarm

user@host> show chassis alarms
1 alarms currently active
Alarm time Class Description
2020-01-14 04:57:44 UTC Minor FPC 2 Minor Errors - CPU Error code: 0x280002

​Logs

Jun 14 04:48:12  alarmd[39959]: Alarm set: FPC color=RED, class=CHASSIS, reason=FPC 2 Minor Errors - CPU Error code: 0x280002
Jun 14 04:58:12  alarmd[39959]: Alarm set: FPC color=RED, class=CHASSIS, reason=FPC 2 Minor Errors - CPU Error code: 0x280002
Jan 14 04:58:13  last message repeated 15 times
Jan 14 04:58:13  last message repeated 8 times
Jan 14 04:58:14  chassisd[3858]: %DAEMON-3:  ASIC Error detected errorno 0x00280002

Explanation

The CPU errors may be generated due to memory errors on the ASIC. There are two types of errors, correctable memory errors, and uncorrectable errors. These errors generate syslog and alarms as follows:

  • Uncorrectable error - shows as "ASIC Error detected errorno 0x280001". This raises a "Major" alarm and syslog
  • Correctable errors have an error rate of less than 255 per 24 hours shows as "ASIC Error detected errorno 0x280002" in syslog and a "Minor" alarm.
  • Correctable memory errors have an error rate greater than or equal to 255 per 24 hours shown as "ASIC Error detected errorno 0x280003" in syslog and a "Major" alarm.

Solution

The correctable memory errors alarm is cleared automatically when the error stops. There is no need to reboot the FPC to clear the alarm. In some cases, the FPC may reboot to clear the alarm on its own. See below for how to check for the history of correctable memory errors.

user@host>show chassis fpc detail

Slot 2 information:
  State                               Online
  Temperature                      29
  Total CPU DRAM                 3200 MB
  Total RLDRAM                    516 MB
  Total DDR DRAM                 8192 MB
  Start time                          2020-06-09 21:59:45 UTC 
  Uptime                              6 hours, 12 minutes, 55 seconds             
  Max power consumption            465 Watts

Steps

  1. Check the FPC alarm count and the active alarm

user@host> show chassis fpc errors        >> Total FPC Minor/Major/Fatal summary
FPC  Level Occurred Cleared Threshold Action-Taken Action
<snip>
2   Minor      38      38     10      38  LOG|
    Major      4       4        1     4   GET STATE|CM ALARM|
    Fatal      0       0        1     0   RESET

user@host> show chassis errors active detail fpc-slot 2
Slot 2

--------------------------------------------------
Location : FPC 2
Identifier : /fpc/2/pfe/0/cm/0/CPU_Error/0/CPU_CMERROR_DDR_CORRECTABLE_MINOR
Error : CPU_CMERROR_DDR_CORRECTABLE_MINOR
Scope : board
Category : processing
Severity : Minor
Details : Minor CPU DDR correctable error
Count : 38
Support : No help info provided
  1. Check for the FPC alarm module.

user@host> request pfe execute command “show cmerror module” target fpc2

 This output will list a number of modules with a module number for each cmerror. Make a note of the module number.

Example Output

<snip>
Module (4) (CPU Error)        >>>>> Make a note of the module number.

Error-id  PFE  Level  Threshold  Count  Occurred  Cleared Last-occurred(ms ago)  Description
--------------------------------------------------------------------------------------------

0x280001 0 Major 1 0 0 0 0        CPU_CMERROR_DDR_UNCORRECTABLE
0x280002 0 Minor 1 0 2 2 28646703 Minor CPU DDR correctable error
0x280003 0 Major 1 0 0 0 0        Major CPU DDR correctable error
<snip>

The error module for the FPC, in this case, is 4. 

  1. Check the error information to determine the reason.

user@host> request pfe execute command "show cmerror module 4 error 0x280002" target fpc2  >> Input error code as seen in alarm

Description           : Minor CPU DDR correctable error
PFE                   : 0
State                 : enabled
Configured Level      : Minor
Default Level         : Minor
Count                 : 0
Threshold             : 1
Error Limit           : 0
Occur Count           : 2
Clear Count           : 2
Last-occurred(ms ago) : 28646703

Logs

----------------------------------------------------------
Index  Time                 Sub-Err   State    Description
----------------------------------------------------------
0      02/04/20 21:00:06    0         Clear    Clear CPU DDR correctable minor alarm
1      02/04/20 20:59:36    0         Set      Raise CPU DDR correctable minor alarm
2      02/04/20 20:54:06    0         Clear    Clear CPU DDR correctable minor alarm
3      02/04/20 20:53:36    0         Set      Raise CPU DDR correctable minor alarm
4      02/04/20 17:40:58    0         Clear    Clear CPU DDR correctable minor alarm
5      02/04/20 17:40:28    0         Set      Raise CPU DDR correctable minor alarm

user@host> request pfe execute command "show cmerror module 4 error 0x280003" target fpc2  >> Input error code as seen in alarm

Description           : Major CPU DDR correctable error
PFE                   : 0
State                 : enable
Configured Level      : Major
Default Level         : Major
<snip>

Since the cause is a hardware memory error, corrected errors are transparent to the system. If the message/alarm occurs just once or very sporadically, it can be ignored.

If the FPC is offline due to the error, there is a possibility that this error is fatal in nature. To recover from this error, ideally, the FPC resets on its own. However, if the FPC alarm has not auto-cleared, you can manually restart the FPC, as shown below.

  1. Restart the FPC if the FPC is offline

user@host> request chassis fpc (offline | online | restart) slot slot-number

In this example:

user@host> request chassis fpc restart slot 2
  1. If a restart does not solve the issue or clear the alarm, contact Support for detailed analysis. This could be a possible hardware issue.

Collect the following logs for troubleshooting and analysis, in addition to the ones collected above:

show log messages
show log chassisd
request execute pfe command "show nvram" target fpc<FPC#>
request execute pfe command "show syslog messages" target fpc<FPC#>

Note: 
For line cards MX2K-MPC7/8/9, the default threshold for single-bit correctable ECC errors on PMB DDR memory is increased from 1 to 10 before declaring a minor alarm. Refer to PR1539779 for enhancement details.

Modification History

2024-05-28: Added clarification about detection interval of 24 hours for 0x280002 and 0x280003; minor formatting changes.
2023-06-26: Added note on enhancement of threshold increased to 10. 
2022-12-27: Fixing typo and reformat
2022-11-04: Update to add condition 0x280001. Clarify that "reboot" is not required for errorno 0x280002 and 0x280003. The alarm will be clear after the error stop for 24 hours.
2022-06-27: Checked article based on feedback and made minor modifications; article found to be valid and accurate
2023-12-25: explanation and content is valid for PTX/QFX boxes as well.

Related Information

[MX] How to disable a specific CMERROR alarm via configuration