[RPKI] [EXTERNAL] routinator 0.10.2 vs. 0.11.2

[Martin Hoffmann]

Havard Eidnes wrote:
> 
> I didn't think I had any overload situations or "slow
> network" locally, but it's obviously not easy to tell where the
> packet loss happened.

Might just be a slow network? Although 300 seconds for 300 Mbytes
should be enough. 

>  Cursory traceroute testing doesn't point
> to a particular problem.  For now I've added
> 
> rsync-timeout = 600
> 
> to my routinator.conf (parameter not in example config, and
> there's no routinator.conf(5) man page, so this is patterned
> after what routinator(1) lists as options).

routinator(1) has a description of all config values as well further
down. We figured having everything in one place is more convenient.

> 
> > What's a bit weird, though, is that it doesn't use RRDP for ARIN.  
> 
> I wonder if that's because the arin.tal lists rsync first, while
> all the others list https (rrdp) first?

No, it prefers HTTPS over rsync if available. And even if it rsyncs the
trust anchor certificate, it should switch to RRDP for the actual
repository if a pointer is provided in the certificate.

> > Can you check your log if there are any errors? There should
> > be.  
> 
> I can't find an rsync error related to ARIN.

Sorry, I meant RRDP errors, specifically anything related to
rrdp.arin.net.

Looks like there indeed isn’t a message logged when it
kills the rsync process. I’ll fix that. 

 -- Martin