03-09-2012 09:52 AM
We have a EX8208 VC in full mesh (two EX8208 witch dual SRE320, and two XRE200) at the core layer and a EX6208 at the access layer (with a LAG).
We have configured these commands in order to have high avalibility on routing engines:
set system commit synchronize
set chassis redundancy graceful-switchover
set routing-options nonstop-routing
In order to test our availability, we have PCs on both sides (core and acces layer) running a ping between them and making a telnet session to the remote switch.
Our tests are:
- We remove a link of the LAG = it works fine! (good).
- We remove from the chassis one SRE320 (EX8208) or SRE64-4XS (EX8208) (which is in master role) = We can see that 5 to 7 pings are lost, and the telnet session is broken (bad ofr us). Ping responds again and we can make a new telnet session.
- We swithover from master to backup routing engine with "request chassis routing-engine master switch" command = We see the same result... lost of pings and telnet session. We can recover connections after 5 secs more less.
Does we need another command to don't loss connections?
03-09-2012 10:55 AM
what version of code are you running?
I am in the process of testing an EX8200-VC configuration and have not seen problems like that with an internal RE failover. However I am usnig EX8216s so the SFs are separate.
also, is STP in use in this environment? Non-stop bridging was not added until 11.3, but I don't think that should come into play with an internal RE failure in an EX8200-VC environment.
03-09-2012 11:03 AM
We have configured NSB directly on members (8208).