hannes changed the topic of #mirage to: MirageOS are OCaml unikernels - https://mirage.io - this channel is logged at http://irclog.whitequark.org/mirage/ - MirageOS 3.7.1 is released - happy hacking!
mort___88 has joined #mirage
mort___8 has quit [Ping timeout: 265 seconds]
noddy has quit [Ping timeout: 256 seconds]
Haudegen has joined #mirage
sora has joined #mirage
zofu has joined #mirage
sora has quit [Ping timeout: 250 seconds]
<zofu> hannes: no restart of the m-fw vm involved. no need to add persistent storage for NAT/fw state.
<zofu> hannes: but it seems to be discarding (and corrupting) all state for all interfaces every time an interface gets added. this is mildly annoying for long-running connections like irc, ssh ...
<zofu> (and right now i cant post it on github because for some reason i can not connect to github anymore and i dont want to restart all the things 27 more times at the moment...)
<zofu> additional question is ... is the mirage-fw supposed to be "stateful"? then thats broken currently too because it starts natting tcp connections that violate the state diagram hard.
<hannes> zofu: there's no state tracking in NAT afair
<hannes> in the OCaml NAT library is what I meant
<zofu> and the rest of the m-fw isnt doing state tracking either?
<hannes> i don't think so
<hannes> and thanks for your reply, marek said similar. i'll look into this with steffi (there must be some code that resets the nat tables then).
<hannes> zofu: has there been state tracking with earlier m-fw?
<zofu> i am not sure.
<zofu> it is possible that the behavior i interpreted as state tracking is really just packets from the "outside" bouncing because of lack of NAT entry.
<zofu> (if you restart a m-fw hard, for the first few minutes it will spam a lot of "waha, dropping this packet on the floor!" for connections that were established before the restart. that is very ok.)
<hannes> (I generally agree that state tracking would be good to have)
<hannes> but I would as well be in a more happy state to have this rather big PR upstreamed before adding more new features ;)
<zofu> agreed. and i am still ok with merging it as-is, as long as we move on to getting the nat-table-reset into an open issue right away.
mezu has joined #mirage
<mezu> (that reconnect was caused by starting a buildvm for testing the fix. ;)
zofu has quit [Ping timeout: 258 seconds]
noddy has joined #mirage
adhux0x0f0x3f has joined #mirage
kuon has joined #mirage
mezu has quit [Ping timeout: 264 seconds]
Haudegen has quit [Quit: Bin weg.]
jiji has joined #mirage
kuon has quit [Ping timeout: 240 seconds]
<jiji> switched main mfw back to the 20200318 build for now, can still test with some that have less impact.
jiji has quit [Remote host closed the connection]
Haudegen has joined #mirage
Haudegen has quit [Quit: Bin weg.]
Haudegen has joined #mirage
Haudegen has quit [Quit: Bin weg.]
Haudegen has joined #mirage
adhux0x0f0x3f has quit [Remote host closed the connection]
adhux0x0f0x3f has joined #mirage
jnavila has joined #mirage
vesper has joined #mirage
vesper11 has quit [Ping timeout: 256 seconds]
Hrundi_V_Bakshi has joined #mirage
aedc has quit [Quit: Leaving]
jnavila has quit [Ping timeout: 256 seconds]
Hrundi_V_Bakshi_ has joined #mirage
Hrundi_V_Bakshi_ has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
Hrundi_V_Bakshi has quit [Ping timeout: 258 seconds]
Haudegen has quit [Ping timeout: 256 seconds]