[j-nsp] Failure detection on Ge interfaces

[Josef Buchsteiner]

    Filippo,

             [ last question I promise]
            
             Can   you   illustrate   to  me  how  you  performed  the
             check/measurement that the link down is not interrupt
             driven?

             The reason I'm interested in this one is that we recently
             fixed a problem on the 1 or 2port GE SFP PIC or the IQ GE
             PIC where the link down event was done in polling cycles
             which has been fixed in PR62682 and is available in
             versions 7.3R3 7.2R3 7.3R2 7.1R4 and higher.
             

            
            thanks
            josef

Monday, October 3, 2005, 5:25:28 PM, you wrote:
FC>    
FC>    
FC> Josef,
FC>  we have PE-1GE-LX-B cards  (1 port Gigabit Ethernet PIC, LX Optics (1310nm,
FC>  single mode fiber) with PIC ejector)
FC>  Thanx
FC>    Filippo
FC>  
FC>  
FC>  
FC>  ----- Original Message -----
FC>  From: "Josef Buchsteiner" <josefb at juniper.net>
FC>  To: "Filippo Cugini" <filippo.cugini at cnit.it>
FC>  Cc: <juniper-nsp at puck.nether.net>
FC>  Sent: Monday, October 03, 2005 4:57 PM
FC>  Subject: Re: [j-nsp] Failure detection on Ge interfaces
FC>  
FC>  
 >> Filippo,
 >>        is the a PIC with SFP connectors ?
 >>        thanks
 >>        Josef
 >>
 >> Monday, October 3, 2005, 2:38:24 PM, you wrote:
 >>
 >> FC>
 >> FC>
 >> FC> Hi all,
 >> FC>  we have M10 with GE LX optical interfaces running ospf and mpls
 >> (junos 7.1).
 >> FC>  We evaluated the M10 behavior in case of failure (fiber cut)
 >> FC>  We have noticed that routers check the operative status of the GE
 >> interface
 >> FC>  once a second, thus determining a significant failure detection time
 >> (from
 >> FC>  0,1 to 1 sec).
 >> FC>  Is it possible to force the router to check the status more
 >> frequently?
 >> FC>  Thanx in advance
 >> FC>  Filippo
FC>   
FC>   
FC>