1
edit
Changes
how to access hardware node from container when hardware node uses routing polisy
Note that every new non-standard gateway will require another table number.
It is likely that your hardware node uses same table to be accessible via both links, not just the link where it's default gateway points. In this case hardware node will not be able to communicate with container. To fix that one needs to add this route into table $TBL
<pre>
# /sbin/ip route add table $TBL "$IP" dev venet0 scope link
</pre>
The same routes are added into 'main' table by vzctl set <CTID> --ipadd <$IP>.
Similar routes should be added for containers to communicate each other.
For more details on routing rules, see <code>man ip</code>.