What's wrong with routes or why is linux ignoring the default gateway?

Question

System administration

Hello Lord!
Interested in this question.
There are two gateways on pfsense 2.0.3 (old version) Let's
call them gw1 and gw2 .
On gw1, vpn ipsec hangs through the provider's vpn (vpn l2 service), on gw2 , vpn hangs through realnik. These vpn from each of the gateways come to Mikrotik (for example). If one provider becomes unavailable, the second one is cut.
Tunnels rise, everything starts to work and respond, except for some important places.
For example, server 88.5 and server 88.55.
I'm starting to figure out what happened. I go to them and do a trace and see that they have a response to my ping through the wrong gateway, although the default gateway is cisco 2801, which pedals all routes.
Routes are added on it as needed. Actually, if other servers can answer me, then I discard the option with the wrong route on the tsiska.
There is a variant with the routing table of servers 88.5 and 88.55
I look at the table in it everything is all right. I look route -C - table cache, it's empty.
then I do ip route flash table all and reboot. About a miracle, everything began to ping and telnet.
Actually the question is, what the hell. Perhaps there are some features in Linux that I don't know about in terms of routing tables? Why and where does it store or cache routes. And why does he not update them himself.
I want to add that at first I tried just to reboot the server and nothing. But as soon as I registered ip route flash + rebooted, everything became as it should.

Answer 1

if this is not some kind of hung session with the old route, then it looks more like a bug in a specific server distribution