<div dir="auto"><div><br><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, 2 May 2023, 09:01 Lincoln Dale, <<a href="mailto:ltd@interlink.com.au" rel="noreferrer noreferrer" target="_blank">ltd@interlink.com.au</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr">All of what MMC said. :)<div>I'd also add: check the prefixes you're announcing via various looking glasses - Vocus and TPG both have ones, but you can go wider too and see what others view you as too. Some nice visualizations from tools like <a href="http://bgp.he.net" rel="noreferrer noreferrer noreferrer" target="_blank">bgp.he.net</a> where you can see their upstreams and how connectivity might work.</div><div><br></div><div>Remember that in routing, the longest prefix always wins, and BGP path selection doesn't even come into that.</div><div>Make sure that your upstreams are accepting your announcements. You cannot announce a more-specific than a /24 and have the world accept it, but it might be that the two /24s you're announcing aren't accepted either, depending on what IRR policies are. You can at least verify that.</div><div>(I'd have done that but don't know your prefixes or ASNs as you didn't post them.)<br></div><div><br></div><div>BGP route propagation does not take 10-15 minutes for an update or withdraw to happen. It's way way quicker than that.</div></div></div></blockquote></div></div><div dir="auto"><br></div><div dir="auto">10 to 15 minute delays sounds like route damping to me.</div><div dir="auto"><br></div><div dir="auto">RIPE recommended disabling it quite a while ago because it had become more trouble than it was worth, and the limits on CPU processing capacity had disappeared as control plane CPUs had become much faster.</div><div dir="auto"><br></div><div dir="auto">However, it seems they've revised that advice with some more suitable damping parameters to make it useful again.</div><div dir="auto"><br></div><div dir="auto"><a href="https://www.ripe.net/publications/docs/ripe-580">https://www.ripe.net/publications/docs/ripe-580</a><br></div><div dir="auto"><br></div><div dir="auto">Regards,</div><div dir="auto">Mark.</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div dir="ltr"><div><br></div><div><br></div><div>cheers,</div><div><br></div><div>lincoln.</div><div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, May 2, 2023 at 8:51 AM Matthew Moyle-Croft <<a href="mailto:mmc@mmc.com.au" rel="noreferrer noreferrer noreferrer" target="_blank">mmc@mmc.com.au</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Have to remember some BGP basics:<div><br></div><div>1) longest prefix (eg. /24 in your case) will always win.</div><div>2) localpref will always win when comparing identical prefixes.</div><div>3) A network will always use localpref to prefer directly connected customer routes.</div><div>4) ASPath length is not going to overcome the above.</div><div><br></div><div>What does "failover" mean to you? When there's a failure, look at what Vocus and TPG have in their route tables and the timing. Also check, are you actually withdrawing the routes during failure?</div><div><br></div><div>MMC </div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, 1 May 2023 at 18:09, Steven Waite <<a href="mailto:steven@waites.com.au" rel="noreferrer noreferrer noreferrer" target="_blank">steven@waites.com.au</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Good evening<br>
<br>
I hope everyone is well. We have a /23 block broken up between TPG /24 and Vocus /24 with the /23 advertise to both Vocus and TPG for failover. This worked will until recently as we noticed increasing failover times during maintenance and now takes around 10-15 minutes. Today I decided to try AS path prepending away from smallest prefix wins type of approach. I think Vocus and TPG ignores prepending as these are local routes thus the local route is preferred even with a lot of prepends. I would love to achieve the same thing via communities if it’s possible. Is someone able to share communities numbers that I should be using for Vocus/TPG please to advertise the primary route for a prefix?<br>
<br>
Many thanks Steve<br></blockquote></div>
</blockquote></div></div>
_______________________________________________<br>
AusNOG mailing list<br>
<a href="mailto:AusNOG@lists.ausnog.net" rel="noreferrer noreferrer noreferrer" target="_blank">AusNOG@lists.ausnog.net</a><br>
<a href="https://lists.ausnog.net/mailman/listinfo/ausnog" rel="noreferrer noreferrer noreferrer noreferrer" target="_blank">https://lists.ausnog.net/mailman/listinfo/ausnog</a><br>
</blockquote></div></div></div>