[j-nsp] 'request system snapshot' causing RPD failures on 7.2R1.3 M20

[Roberts, Michael J. (IATS)]

I am having some trouble with an M20 in our lab.  I am trying to perform
'request system snapshot' on the RE0 and RE1.  

If I perform the command on the master RE, the router crashes.  If I
connect to the backup, and perform the command, it works without a
restart.  Graceful restart is configured, and interesting enough, it is
not working.  I have a full blown 90 second outage on the master RE when
this occurs.  The backup RE is not taking over.

Here is what is written to the messages on the master when the snapshot
is performed:

Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: tnp_hello_input Received TNP
server hello when we were master!
Jun 14 08:46:05  LSBTESTR1_RE0 rshd[3725]: root at re1 as root: cmd='rcp -T
-t /var/db/lmpd-name-id.db'
Jun 14 08:46:05  LSBTESTR1_RE0 ssb CM(0): Routing engine CM reconnection
succeeded after 0 tries
Jun 14 08:46:05  LSBTESTR1_RE0 ssb CM(0): ALARM CLEAR: RE chassis socket
closed abruptly
Jun 14 08:46:05  LSBTESTR1_RE0 fpc2 PFEMAN master RE reconnection made
Jun 14 08:46:05  LSBTESTR1_RE0 fpc0 PFEMAN master RE reconnection made
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_peer_alloc: Connection
request when there is a pending connection for peer index 2, type 3
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_listener_connect: connect
from tnp 0x12: too many open connections
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_peer_alloc: Connection
request when there is a pending connection for peer index 0, type 3
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_listener_connect: connect
from tnp 0x10: too many open connections
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_peer_alloc: Connection
request when there is a pending connection for peer index 2, type 3
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_listener_connect: connect
from tnp 0x12: too many open connections
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_peer_alloc: Connection
request when there is a pending connection for peer index 0, type 3
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: pfe_listener_connect: connect
from tnp 0x10: too many open connections
Jun 14 08:46:05  LSBTESTR1_RE0 /kernel: tnp_hello_input Received TNP
server hello when we were master!
Jun 14 08:46:06  LSBTESTR1_RE0 fpc2 PFEMAN master RE reconnection made
Jun 14 08:46:06  LSBTESTR1_RE0 ssb RDP: Remote side reset connection:
rdp.(scb:26626).(serverRouter:chassis)
Jun 14 08:46:06  LSBTESTR1_RE0 fpc2 RDP: Remote side closed connection:
rdp.(fpc2:6147).(serverRouter:pfe)
Jun 14 08:46:06  LSBTESTR1_RE0 ssb CM(0): ALARM SET: (Major) RE chassis
socket closed abruptly
Jun 14 08:46:06  LSBTESTR1_RE0 /kernel: pfe_peer_alloc: Connection
request when there is a pending connection for peer index 0, type 1


There is more info, but I did not want to flood everyone's inbox.  Our
maintenance is currently being renewed and changed over to Juniper, so I
am not sure if I can even open a case at this moment, but I wanted to
check with you all and see if anyone has any suggestions.  Thx.

-mike