This may not always be an option, but I would use tagging and route-maps instead of distribute-lists any time I can. There is just too much room for serious, show-stopping errors, with distribute-lists. With route tagging, you could mess up and end up with a packet taking a suboptimal path to a destination, and you could lose some points, but the packet will get to its destination and whatever was riding on this will probably work.
But that's just me working around my natural absent-mindedness-- taking as a given that I will screw up the distribute lists. Tom Larus, CCIE #10,014 ""The Long and Winding Road"" wrote in message news:[EMAIL PROTECTED] > For the past couple of weeks I've been whacking out various CCIE practice > labs. I've also been suffering various degrees of euphoria and depression, > depending upon how badly I was suckered by the redistribution problems. > > After a particularly long and frustrating day with the Cisco ASET Lab #1, it > suddenly occurred to me that there are many ways to do things, and for some > reason, I've been overlooking what may be the best way to deal with > redistribution. > > Those of you who have worked these practice labs know how it goes. You read > through the lab, then you start configuring. > > Step 1 - set up OSPF > Step 2 - set up RIP > Step 3 - redistribute between OSPF and RIP > Step 4 - set up EIGRP > Step 5 - redistribute between EIGRP and RIP > Step 6 - set up IS-IS > Step 7 - redistribute between IS-IS and OSPF > Step 8 - scream in anguish as you discover that your routing tables have > turned to trash and half your network becomes unreachable. > > ASET #1 was particularly nasty in how it accomplished Step 8 > > Which brings me to the topic of this post. CCIE's and folks who've been > through the Lab without success - what do you think of this approach: > > 1) do NOT do any redistribution anyplace until all routing protocols have > been configured everywhere. Yes, I know that typically you have a section > with several steps, one of which is redistribution. But mark your place and > return after the IGPs are up and running and all routes for a particular IGP > are where they should be. > > 2) return to the first redistribution task. Before configuring anything, > refer to your diagram ( you DO write out a nice diagram, don't you? ) and > ask yourself: "after I do one way redistribution, what routes will appear > where?" > > 2a) Consider how administrative distance might change things > > 2b) Follow the redistribution to it's extreme. For example, if you > redistribute EIGRP into OSPF, what routers will these routes end up on? Will > there be any implications to the routing tables? > > 3) repeat step 2 for every redistribution point, each time considering the > totality of the contents of the redistributed routes. So if you have > redistributed IS-IS into OSPF, how do those redistributed routes flow > through the OSPF domain? > > 4) Keep an eye out for things like split horizon > > 5) every step along the way, consider what routers need to see what routes. > Watch for situations where necessary routes do not appear. ( you have > probably trashed it because of overzealous filtering. ) > > 5) If problems occur, such as a routing loop, trace back where the problem > route came from, and see what can be done to evade the problem. Summary > routes work wonders sometimes. So do route-maps and distribute lists. > > Re-reading this, I see that this topic does not lend itself well to text. I > can say with certainty that I now have a very clear vision of redistribution > methodology. I've tested it three times now with different labs, and I > believe I am solving the redistribution problems more quickly than ever. I > hope that I have painted enough of a picture that some of you can fill in > the rest. > > Chuck > > > -- > TANSTAAFL > "there ain't no such thing as a free lunch" Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=66343&t=66306 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
