[j-nsp] juniper-nsp Digest, Vol 93, Issue 40

[Haider Ali]

HI Friends
I have to implement SRX650 in cluster but facing issues regarding that. I was using JUNOS 10.0R3.10 on SRX650 cluster, but faced multiple issue like, kernel shooting to 95%, many of the services were not working properly, unable to see the routes in the routing table, although the next hop is reachable and active at that time, core-dump was generated by SRX naming  “flowd_octeon_hm”. When I saw the output of “show system process summary”, I can see there is a process naming “flowd_octeon_hm” using around 1039% of the WCPU. I have opened a ticket to Juniper Networks and they suggested me to move to JUNOS 10.1R3.7, as this core-dump was generated due to a bug inside the JUNOS 10.0R3.10. But after upgrading the JUNOS to 10.1R3.7, I can see the same process using the same amount of WCPU, although the boxes are not in production, so no traffic is there on the boxes, but the WCPU still using the same amount of memory. Latterly when I told JTAC regarding this, they found that its caused due to a bug inside the recommended JUNOS also. So they said to move to Junos 10.2. I just want to know if anyone have observed this kind of behavior of SRX650 in cluster ? If yes, please share the procedure and steps which you implemented to get this problem resolved.
 
Concerns:
1)    FTP service was not working properly on SRX650, although the ALG was disabled. There will be 800+ FTP session when the boxes will be placed in production.
2)    Kernel shoots up to 95% with minimal amount of traffic on SRX650 Cluster (traffic around 25-30 MB)
3)    SRX-650 Cluster not passing the policy traffic logs to NSM-2010.2
4)    I have to enable all the logging on the boxes, the IDP logging and the policy logging.
 
NOTE: Please suggest the JUNOS version which caters the resolution of all the above mentioned issues.
 
Please share your comments on this.

Regards

Haider Ali
0092-331-2827010