This KB describes the scenario when continous generation of parity error is seen and the way to troubleshoot the same
Continous generation of below parity error logs will be seen on the device :-
May 2 05:00:01 device : %PFE-3: fpc0 Unit: 0
May 2 05:00:01 device : %PFE-3: fpc0 Mem:
May 2 05:00:01 device : %PFE-3: fpc0 Parity error..
May 2 05:00:01 device : %PFE-3: fpc0 Error in: SBUS transaction.
May 2 05:00:01 device : %PFE-3: fpc0 Blk: 1, Pipe: 1, Address: 0x241026e5, base: 0x4, stage: 9, index: 9957
May 2 05:00:01 device : %PFE-3: fpc0 Unit 0: mem: 2095=L3_ENTRY_IPV4_UNICAST blkoffset:10
May 2 05:00:01 device : %PFE-3: fpc0 Unit 0: _soc_ser_overlay_mem_correction: mem write L3_ENTRY_ONLY.ipipe0[9957] failed: Operation failed
This issue happens on QFX5k devices running a Junos release which doesnot supports auto detection and correction of parity . Along with that if the device uptime is very high then this issue is likely to occur.
Below steps can be followed to mitigate this issue :-
>>Collect below Output from the device to check the counter for Parity error :-
request pfe execute command "set dcbcm bcmshell 'soc'" target fpc0
Sample Output in problem state :-
user@device> request pfe execute command "set dcbcm bcmshell 'soc'" target fpc0
<truncated output>
SENT: Ukern command: set dcbcm bcmshell 'soc'
GOT: HW (unit 0)
GOT: Unit 0 Driver Control Structure:
GOT: Chip=BCM56850_A2 Rev=0x03 Driver=BCM56850_A0
GOT: Error: SDRAM=0 CFAP=0 Fcell=0 MmuSR=0
GOT: SER events(mem=244933 reg=0 nak=0 stat=0 ecc=0 direct=0 fifo=244932 tcam=0)
GOT: SER corrections(fix=2 clear=16 restore=0 special=0 err:0)
>> Reboot the device
>> Collect same Output again :-
GOT: SER events(mem=0 reg=0 nak=0 stat=0 ecc=0 direct=0 fifo=0 tcam=0)
GOT: SER corrections(fix=0 clear=0 restore=0 special=0 err:0)
The device should be monitored for 24 hours and if the same Parity error logs are seen again then contact JTAC for further support.