[j-nsp] Odd behaviour testing failover

[Josef Buchsteiner]

Thursday, September 8, 2005, 4:52:50 AM, you wrote:

GS>    
GS>    
GS> Hi all,
GS>  
GS>  I'm seeing odd RE behaviour on an M20 while testing RE failover.
 >>From the log entries, I'm guessing that a snapshot of the current route
GS>  table entries gets tarballed and passed to the second RE

     tarball creation  stays local which is done by dumpd. however you
     had a rpd core which does invoke dump to create a tarball file
     with relevant informations like adding config and message log to
     the core file..

GS>  to use while
GS>  the control plane re-establishes adjacencies with other devices.
GS>  
GS>  Problem is, the tarball can't be found by the RE, and transit traffic
GS>  fails until it re-meshes.


           dumpd creates the tar file and rpd does write the core.
           please look if you still have a rpd.core file in the
           /var/tmp/ directory generated from rpd.
GS>  
GS>  Has anyone come across this before?
GS>  If so, how do I get this to behave? Graceful failover is enabled on this
GS>  box (JUNOS 7.3R1.6)
GS>  
GS>  Also seeing kernel page faults in the logs on the backup RE during
GS>  failover. Hard disk failing?


     this is a rpd page fault.. again you should have a rpd.core file
     please issue a case with all the data so we can fix this

     thanks
     Josef


GS>  
GS>  
GS>  
GS>  Sep  8 14:36:35  jcore2 chassisd[74828]: CHASSISD_SNMP_TRAP7: SNMP trap
GS>  generated: Fru Online (jnxFruContentsIndex 9, jnxFruL1Index 2,
GS>  jnxFruL2Index 0, jnxFruL3Index 0, jnxFruName Routing Engine 1,
GS>  jnxFruType 6, jnxFruSlot 2)
GS>  Sep  8 14:36:35  jcore2 craftd[2599]: attempt to delete alarm not in
GS>  list
GS>  Sep  8 14:36:35  jcore2 craftd[2599]: forwarding display request to
GS>  chassisd: type = 4, subtype = 44
GS>  Sep  8 14:36:41  jcore2 rshd[75266]: root at re1 as root: cmd='rcp -T -t
GS>  /var/db/dcd.snmp_ix+'
GS>  Sep  8 14:36:42  jcore2 rshd[75298]: root at re1 as root: cmd='mv
GS>  /var/db/dcd.snmp_ix+ /var/db/dcd.snmp_ix'
GS>  Sep  8 14:36:59  jcore2 dumpd: Core and context for rpd saved in
GS>  /var/tmp/rpd.core-tarball.4.tgz
GS>  Sep  8 14:37:00  jcore2 dumpd: tar: rpd.info.4: Cannot stat: No such
GS>  file or directory tar: Error exit delayed from previous errors
GS>  Sep  8 14:37:00  jcore2 dumpd: Unable to create core tarball
GS>  /var/tmp/rpd.core-tarball.4.tgz
GS>  Sep  8 14:37:00  jcore2 dumpd: tar: rpd.info.4: Cannot stat: No such
GS>  file or directory tar: Error exit delayed from previous errors
GS>  Sep  8 14:37:00  jcore2 dumpd: Unable to create core tarball
GS>  /var/tmp/rpd.core-tarball.4.tgz
GS>  
GS>  
GS>  
GS>  Sep  8 14:36:24  jcore2 /kernel: Trapframe Register Dump:
GS>  Sep  8 14:36:24  jcore2 /kernel: eax: 00000000  ecx: 085cf000   edx:
GS>  085fabb0   ebx: 00000012
GS>  Sep  8 14:36:24  jcore2 /kernel: esp: bfbff800  ebp: bfbffc28   esi:
GS>  085c63a0   edi: 0864c000
GS>  Sep  8 14:36:24  jcore2 /kernel: eip: 0812f706  eflags: 00010206
GS>  Sep  8 14:36:24  jcore2 /kernel: cs: 001f       ss: 002f        ds:
GS>  bfbf002f    es: 8868002f
GS>  Sep  8 14:36:24  jcore2 /kernel: fs: 864002f    trapno: 0000000c
GS>  err: 00000004
GS>  Sep  8 14:36:25  jcore2 /kernel: Page table info for PC address
GS>  0x812f706: PDE = 0x189a5067, PTE = 2c65425
GS>  Sep  8 14:36:25  jcore2 /kernel: Dumping 16 bytes starting at PC address
GS>  0x812f706:
GS>  Sep  8 14:36:25  jcore2 /kernel: 80 b8 10 02 00 00 00 75 0d 80 bd e7 fb
GS>  ff ff 03
GS>  Sep  8 14:36:25  jcore2 /kernel: BAD_PAGE_FAULT: pid 62594 (rpd), uid 0:
GS>  pc 0x812f706 got a read fault at 0x210, x86 fault flags = 0x4
GS>  
GS>  
GS>  _______________________________________________
GS>  juniper-nsp mailing list juniper-nsp at puck.nether.net
GS>  http://puck.nether.net/mailman/listinfo/juniper-nsp
GS>   
GS>   
GS>