iroute 0.0.0.0 not work and not warning about that

[mastervolkov]

If in ccd (for cl5, for example) set:
iroute 0.0.0.0 128.0.0.0 192.168.5.5
iroute 0.0.0.0 128.0.0.0 192.168.5.5

Then in log we view:
: cl05/X.X.X.X:53 MULTI: internal route 128.0.0.0/1 -> exitnode/X.X.X.X:53
: cl05/X.X.X.X:53 MULTI: internal route 0.0.0.0/1 -> exitnode/X.X.X.X:53

, but work as expected only 128.0.0.1 - 255.255.255.255, for example in log we see:

: cl05/Y.Y.Y.Y:53 MULTI: Learn: 128.0.0.1 -> exitnode/X.X.X.X:53

All query to 0.0.0.0 - 127.255.255.255 do not exist in the journal and, accordingly, do not work.

If that normal, is can any warning be except simple "internal route 0.0.0.0/1 -> exitnode/X.X.X.X:53", similary wrong configured subnet, etc

[Gert Döring]

it should work - this is code that was written many years ago, so arguably it might be a bug, or a conflict with internal implementation details, or someone thought it would be reasonable to reject 0.0.0.0 - dunno, someone will need to investigate and fix it.

{{{ iroute-ipv6 2000::/3 }}

should work just fine, though :-)

[mastervolkov]

Temporary fix problem with 0.0.0.0 by add other network

[mastervolkov]

This problem found by other people:

​http://serverfault.com/questions/310002/openvpn-iroute-for-all-theoretically-reachable-public-ips

[Antonio Quartulli]

Apparently mroute_learnable_address() returns FALSE for an address made up by 0s only ... thus multi_learn_addr() refuses to add the route to the internal routing table :-(

mroute_learnable_address() could be changed to accept this case, but this needs to be carefully checked to understand what's the impact on other code.

My guess is that this is just a sanity check before adding a route and a 0/x was not thought to be a meaningful one.

[Antonio Quartulli]

@mastervolkov: a patch (based on current master) addressing this issue has been sent to the mailing list. Its subject is:
"[PATCH] Allow learning iroutes with network made up of all 0s (only if netbits < 8)"

If you are still interested in this problem, would you mind giving it a try and reporting the results here?

Thanks a lot!

[Antonio Quartulli]

The patch fixing this issue has been merged into master (commit a19c56db) and the 2.4 (commit 80614c7c) branches. It'll be available in 2.5 and in the next 2.4.x release.