IRCaBot 2.1.0
GPLv3 © acetone, 2021-2022
#ls2
/2022/03/08
zzz DST is this weekend over here, changing meeting UTC time to keep it the same for us, heads up zlatinb
zlatinb ok thx
zzz I believe you're two weeks later than us
zzz zlatinb, I need a primer on adding delay/drop to testnet
zlatinb zzz: /sbin/tc qdisc add dev eth0 root netem delay 25ms
zlatinb from inside the container
zlatinb man 8 netem
zlatinb when done, either bounce the container or /sbin/tc qdisc del dev eth0 root
zlatinb adding loss has only worked for random type for me, I tried correlation once and it ended up using all my cpu
zzz so I need to do that in each container?
zzz I thought maybe I could just add it to the hub device wlpxxxx but no need to reinvent the wheel here
zlatinb I added it to each container:
zlatinb for i in $(seq 1 15); do lxc-attach i2p$i -- /sbin/tc qdisc add dev eth0 root netem delay 25ms; done
zzz ok, thanks
zlatinb I haven't had to touch the bridge so far, but I've been thinking that to test session migration we may need to do some magic there
zzz what's your standard drop % values you test with?
zlatinb usually I try 0.1, 0.25 and 0.5 when we test performance/flow control changes
zlatinb those are enough to trigger the recovery code when doing an eepget, idk if they'll be enough for idle connections
zzz so it would be loss random 0.25 ? and does that go in the same command with the delay, or a separate tc qdisc add?
zlatinb same command as delay
zlatinb you can't "update" a qdisc, you need to delete the old one then add a new one
zzz ok so delay xxx loss random xxx
zzz got it
zzz got it working, RTT is up and seeing some NACKs
orignal zzz, your banner on zzz.i2p
orignal if I was you I would stay neutral
zzz yeah, good point
orignal but if you on that side than you should change your nick )))
orignal zzz stands for Z
eyedeekay If the operator of dist.torproject.i2p is in this room, please contact me at your earliest convenient time.
R4SAS eyedeekay: ?
eyedeekay I'm having a pretty unusual issue with it
eyedeekay I can't verify the signature when I download from that mirror, but I can verify the signature from every other mirror
eyedeekay I thought it was the download being incomplete, so I added resumability
eyedeekay And now I know the file is complete, still a signature error
R4SAS hm, nothing must be changed, because this is only proxy_pass in nginx
R4SAS but let's try one thing
R4SAS I'll remove all sub_filter's for that endpoint
eyedeekay Let me know when you've done it so I can re-test
R4SAS done, checking myself too
R4SAS locally
eyedeekay Mine takes a while, have to download it over I2P so it'll be a few minutes before I know
R4SAS tested locally, everything fine
R4SAS I see same thing with previously downloaded binary
R4SAS looks like nginx somehow made replaces in binary, but that musnt happen
eyedeekay Thanks for checking it out