[j-nsp] Cluster issue with SRX550

[Per Westerlund]

 From your post it is not obvious what setup you are using. Can you 
elaborate, please?

Two standard setups that normally work:

- One link from each of node0 and node1, to (in this case) MX5-T. No 
LACP, just active/passive setup with RETH-interface in SRX cluster.

- At least two interfaces from each of node0 and node1 to MX5-T. Two 
LAGs, one from node0 to MX5-T and (another) from node1 to MX5-T. LACP 
used (2 aeX at MX5-T), one RETH-interface at SRX cluster.

/Per

On 24 May 2014, at 18:33, Ali Sumsam wrote:

> Hi All,
>
> *Scenario*
> We have a cluster of two SRX550 and a MX5-T router.
> An aggregated link (LACP) is connecting node0 (primary node) of SRX 
> 550
> cluster to MX5-T router.
> The aggregate link consists of two ports. Copper/Ethernet.
>
> The devices are in the same rack and directly connected (in case 
> someone
> doubts the physical connectivity)
>
> *Junos*
> SRX550 Cluster 12.1X44-D30.4
> MX5-T 11.4R7.5
>
> *Problem*
> Sometimes it is observed that the LACP goes down on the MX5-T router. 
> At
> that time following logs are seen.
>
> LACPD_TIMEOUT: ge-1/0/4: lacp current while timer expired current 
> Receive
> State: CURRENT
> /kernel: KERN_LACP_INTF_STATE_CHANGE: lacp_update_state_userspace: new
> state is 0 cifd ge-1/0/4
> /kernel: KERN_LACP_INTF_STATE_CHANGE: lacp_update_state_userspace: new
> state is 0 cifd ge-1/0/5
> /kernel: ae_bundlestate_ifd_change: bundle ae1: bundle IFD minimum 
> links
> not met 0 < 1
>
> At this time, the SRX cluster tries to failover and following logs 
> appear.
>
> jsrpd[1393]: JSRPD_RG_STATE_CHANGE: Redundancy-group 2 transitioned 
> from
> 'secondary' to 'primary' state due to Remote node is in secondary hold
> jsrpd[1393]: JSRPD_RG_STATE_CHANGE: Redundancy-group 1 transitioned 
> from
> 'primary' to 'secondary-hold' state due to Monitor failed: IF
> jsrpd[1393]: JSRPD_RG_STATE_CHANGE: Redundancy-group 2 transitioned 
> from
> 'primary' to 'secondary-hold' state due to Monitor failed: IF
> jsrpd[1393]: JSRPD_RG_STATE_CHANGE: Redundancy-group 1 transitioned 
> from
> 'secondary-hold' to 'secondary' state due to Back to back failover 
> interval
> expired
> jsrpd[1393]: JSRPD_RG_STATE_CHANGE: Redundancy-group 1 transitioned 
> from
> 'secondary' to 'primary' state due to Remote node is in secondary hold
> jsrpd[1393]: JSRPD_RG_STATE_CHANGE: Redundancy-group 2 transitioned 
> from
> 'secondary-hold' to 'secondary' state due to Back to back failover 
> interval
> expired
>
> Minutes later, the node0 comes back as primary and service is 
> restored.
>
> Besides,  following logs on the SRX550 are coming.
>
> /kernel: Process with Most Children- 1:init - Children - 211
> /kernel: maxproc limit exceeded by uid 0, please see tuning(7) and
> login.conf(5).
> /kernel: nearing maxproc limit by uid 0, please see tuning(7) and
> login.conf(5).
> /kernel: Process with Most Children- 1:init - Children - 211
> /kernel: maxproc limit exceeded by uid 0, please see tuning(7) and
> login.conf(5).
> /kernel: Process with Most Children- 1:init - Children - 211
> /kernel: maxproc limit exceeded by uid 0, please see tuning(7) and
> login.conf(5).
> /kernel: Process with Most Children- 1:init - Children - 211
> /kernel: maxproc limit exceeded by uid 0, please see tuning(7) and
> login.conf(5).
> /kernel: nearing maxproc limit by uid 0, please see tuning(7) and
> login.conf(5).
> /kernel: Process with Most Children- 1:init - Children - 211
>
>
>
> We have been recommended to change following
>
> set interfaces reth1 redundant-ether-options lacp periodic fast
>
> to
>
> set interfaces reth1 redundant-ether-options lacp periodic slow
>
>
>
> If someone had similar experience, would appreciate your help.
>
> Regards,
>
> *Ali Sumsam CCIE - *eintellego Networks Pty Ltd
> Senior Network Engineer
> ali at eintellegonetworks.com ; www.eintellegonetworks.com
>
> Phone: 1300 239 038; Cell +61 (0)450 609 592 ; skype://sumsam.ali80
>
> facebook.com/eintellegonetworks ;  <http://twitter.com/networkceoau>
> linkedin.com/in/alisumsam
>
>
> The Experts Who The Experts Call
> Juniper - Cisco - Cloud
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp