Powerconnect switches

[~Darkseeker~]

Yo!

a business related question, unusual for me I know

We've been playing with a fibre line at a client site for a while trying to get a solid connection down it. It's a point to point fibre connection that links two Dell Powerconnect switches together (a main office switch to a satellite office).

In the main cab (the one with the internet connection and such, running on two seperate ADSL lines via Juniper SRX110 routers for redundancy) we have a PowerConnect N2048 switch with a 1gbps Dell fibre tranceiver in it. At the other end, we have a PowerConnect 5524 switch with an identical Fibre tranceiver installed.

At both ends, the Fe/0/1 port (the first fibre port) on each device is set to auto-negotiate speed and has rapid STP disabled (we've had issues in the past running rapid spanning tree on uplink ports)

We get traffic across the link using this method, and DNS at the satellite end is able to resolve the domain name of the network and ping centrally-held off-site servers using their FQDN (which is unusual since we don't normally have to use FQDNs since it's a WAN and usually hostname is enough to get a ping)

However the connection is flakey, connection across the link is unstable and most of the aspects of our managed build that use network connections (group policy for example) don't seem to be working. I've been thinking it could be because the switches at both ends are 3-4 years apart in age, and presumably in that time the fibre technology has changed (for instance, there are a LOT more settings regarding fibre and such on the N series). what do y'all think?

 

[root]

Do you only have 1 piece of fibre between the switches? - no redundancy?

Random drop outs disconnects etc could be a spanning tree issue?
you've disabled rapid spanning tree, but are you still using old slow spanning tree? - have you set root bridge priorities to how you want then to be setup? if spanning tree is regularly changing and needing to reconverge that would look strange fro a connectivity point of view.

otherwise I'd look to hard code negotiate settings for a start to rule out negotiation mismatch. (this sort of thing can happen with one end trying to be 100mb half whilst the other is trying 1gb full duplex.
(at least check, (there should be an equivalent to show int fa1 status? - it might be that?) check at both ends obviously.
- what mini GBIC/SFP are you using, - are the supported as well as compatible?
(you're not doing something like using a 4gb sfp in a 1gb only port?)

Are the SFPs properly matched to the fibre you are using LC vs SC connectors are a given (and SC connector won't fit in an LC hole!) but what wavelength, are there joins in the fibre etc - and were they done properly.
if you're using SC are you sure the (usually black and red) boots are the right way around? - you are not plugging TX - TX and RX - RX?

that you need to use FQDN and one site an not the other is weird. - I assume you're running DHCP and have 1 server that's serving addresses to both sites/buildings. if you're setting manually, or using 2 servers, have you added scope options for DNS search suffix?

 

[~Darkseeker~]

True to form it turns out it was actually a combination of the first two things we checked... we changed the fibre patches and the Dell transceivers on day one... apparently the ones we put in were faulty. After changing them all again and having a rant at dell for sending us faulty equipment twice in a row it seems to be working now... haha

 

[setishock]

How did they explain 2 times sending you faulty gear? I'd bet real money they sent you what was sent back by some else to be refurbished but didn't get that far along the repair chain.