avsm changed the topic of #mirage to: mirage 2 released! party on!
_whitelogger has joined #mirage
justincormack has quit [Read error: Connection reset by peer]
justincormack has joined #mirage
rgrinberg has quit [Remote host closed the connection]
nullcat has joined #mirage
nullcat_ has joined #mirage
nullcat has quit [Ping timeout: 260 seconds]
nullcat_ has quit [Ping timeout: 244 seconds]
nullcat has joined #mirage
noddy has quit [Ping timeout: 256 seconds]
jermar has joined #mirage
noddy has joined #mirage
copy` has quit [Quit: Connection closed for inactivity]
nullcat has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
AltGr has joined #mirage
andreas23 has joined #mirage
fgimenez has joined #mirage
jermar has quit [Read error: Connection reset by peer]
calculemus has joined #mirage
fgimenez has quit [Ping timeout: 256 seconds]
fgimenez has joined #mirage
fgimenez has joined #mirage
fgimenez has quit [Changing host]
jermar has joined #mirage
calculemus has quit [Quit: WeeChat 1.6]
henrytill has quit [Ping timeout: 260 seconds]
calculemus has joined #mirage
fgimenez has quit [Ping timeout: 260 seconds]
fgimenez has joined #mirage
jgjl has joined #mirage
fgimenez has quit [Ping timeout: 256 seconds]
fgimenez has joined #mirage
fgimenez has joined #mirage
fgimenez has quit [Changing host]
djs551 has joined #mirage
djs55 has quit [Ping timeout: 250 seconds]
djs551 has quit [Ping timeout: 260 seconds]
djs55 has joined #mirage
yomimono has joined #mirage
<hannes> AltGr: around? is there a way to get the package download stats (package X downloaded Y times) which does not involve to scrape opam.ocaml.org? and are there cumulative stats (not only the 'this month' stats) over a longer period of time?
rgrinberg has joined #mirage
agarwal1975 has joined #mirage
yomimono has quit [Ping timeout: 244 seconds]
<AltGr> at the moment, stats are generated by opam2web by parsing the raw nginx log
<AltGr> If you have a better tool in mind that could do the parsing and return concise anonymous data, we could put that on the server and make the results available
<AltGr> (I would actually be interested, as I changed the user-agent in opam 2, and could have stats on how many ppl are using it already ; but it's too much work at the moment)
yomimono has joined #mirage
<hannes> AltGr: I'd be interested (to have some evidence for arguments) in a quadruple (month * package * version * count) list.. are the nginx logs thrown away after processing?
<AltGr> No no, we still have them from the start
<hannes> nice. I might look into opam2web and provide you with something to run on the logs...
<AltGr> I'd gladly make the data available, if we can preserve privacy
<hannes> AltGr: that's why I'd prefer you run a shell script which produces data I'd be interested in ;)
<hannes> OTOH it would likely be fair to exclude travis + docker and other automated builds..
<Drup> (please distinguish raw count and unique count)
<AltGr> if you provide it, we can run it on the server; you can even PR to https://github.com/ocaml/opam.ocaml.org actually
<hannes> I currently do some empirical analysis of the opam-repository git, but that doesn't give me the 'popularity' index
<AltGr> note that Anil is currently preparing a migration from the VM to containers
<AltGr> Drup: I think opam2web shows only unique count
<hannes> AltGr: I hope the nginx logs won't be lost during migration
<AltGr> no no, I made sure that's being taken care of
<AltGr> they do contain valuable info
<hannes> \o/
<hannes> AltGr: I also looked through Camelus source, and will hopefully find some time to get something similar regarding signing soon
<AltGr> great
<hannes> (unless I keep on procrastinating with other projects, such as MirageOS3 release)
<hannes> I recently met this justin cappos guy, who is one of the TUF authors, and got motivated again to do some empirical data to justify design decisions
yomimono has quit [Ping timeout: 244 seconds]
<AltGr> cool!
<mato> hannes: Got a minute? ricarkol is telling me that he no longer sees any packet loss with Solo5 latest master on his machine running 12.0-CURRENT.
<mato> hannes: (presumably post the last PR which fixed the IP/ICMP checksum)
<mato> hannes: I still see packet loss on my machine, but that's on 11.0-RELEASE. Can you give it a go on yours?
<hannes> mato: sure, let me find the magic commands
<hannes> 112551 packets transmitted, 110359 packets received, +174 duplicates, 1.9% packet loss
<hannes> [15:44 hannes@aternefas:test_ping_serve] sudo ping -i 0.0001 10.0.0.255
<hannes> round-trip min/avg/max/stddev = 0.016/4.399/15.874/3.966 ms
<hannes> ping -f is worse, though
<hannes> 127959 packets transmitted, 28442 packets received, +211 duplicates, 77.8% packet loss
<hannes> round-trip min/avg/max/stddev = 0.037/16.116/26.961/9.542 ms
<hannes> but to get a reliable ping -f we should likely do some input processing limits..
<mato> ok, that matches what I see, i.e. just as screwy as ever
<mato> Is the Linux ping -f different then?
<mato> hannes: When we tested together in Berlin, host<->guest ping -f with FreeBSD guest was "reliable", wasn't it?
<hannes> mato: it is a big improvement over the previous run (in berlin when we sat together): there ping -d 0.0001 (even -d 0.001 iirc) was >85% loss
<mato> ah, right
<hannes> yes, FreeBSD host _and_ guest had a good ping -f behaviour
<hannes> even ping -i 0.0000000000000000000000000000000000000000001 behaves sanely (3% packet loss)
<mato> hannes: Hmm. Docs for ping -f on Linux and FreeBSD say the same thing (100 times per second or as fast as replies come back, whichever is greater)
<mato> That's... weird.
* mato is wondering if this is a real bug or a wild goose chase
<mato> hannes: how are you invoking bhyve? (exact command line)
yomimono has joined #mirage
<hannes> sudo bhyve -A -H -P -s 0:0,hostbridge -s 1:0,lpc -s 2:0,virtio-net,tap0 -l com1,stdio test
yomimono has quit [Ping timeout: 260 seconds]
<hannes> mato: I would not trust the documentation of ping, also the tap interface might be configured differently... I manually do a ifconfig tap0 create, which then has the following options/flags:
<hannes> tap0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
<hannes> options=80000<LINKSTATE>
andreas23 has quit [Quit: Leaving.]
<hannes> (but I don't know where I'd see things like ring buffer size)
yomimono has joined #mirage
rgrinberg has quit [Ping timeout: 256 seconds]
yomimono has quit [Quit: Leaving]
rgrinberg has joined #mirage
copy` has joined #mirage
smondet has joined #mirage
<mato> hannes: Aha! Progress!
<hannes> mato: where?
<mato> hannes: Can you grab latest master (fixes a stupid bug in the ping test), add a static ARP entry for the bhyve VM and then run ping -f against 10.0.0.2 (*unicast*)?
* mato bashes head against something for relying on broadcast ping behaviour to be sane
<hannes> mato: yes! works!
<hannes> btw, where does solo5 get its mac address from?
<hannes> 62492 packets transmitted, 62491 packets received, 0.0% packet loss
<hannes> round-trip min/avg/max/stddev = 0.024/0.031/0.263/0.014 ms
<hannes> (and yes, I tried earlier to do a unicast ping, but didn't get a reply. thx for fixing all of this!)
<mato> hannes: from bhyve (mac address is part of the virtio spec)
<mato> *party*
<hannes> \o/
<mato> As I told ricardo on Slack, I wouldn't worry about broadcast ping, the behaviour there is somewhat undefined...
<hannes> this is actually good news!
<hannes> next would be to measure some MirageOS app on top of it!?
<mato> yeah, the best thing to do there is to resurrect/update the iperf unikernels
<mato> and run it against iperf from the host
<hannes> this is test code, but not the standalone unikernel... but yeah.. wrapping this as a unikernel would be great to have (maybe add to mirage-skeleton)
<mato> yes. i was going to look at doing that
<hannes> in case you need help there (/with OCaml), feel free to ask
<mato> will do
<hannes> mato: unsure whether you read it or not, but I implemented a syslog over TLS for mirage (reporter from the Logs library, mutual authentication, to be used instead of the console reporting) :)
<mato> hannes: saw the subject, have not read it yet
<hannes> (there's also a udp and a tcp version https://github.com/hannesm/logs-syslog :)
<mato> cool
<hannes> one of the motivations was you mentioning we need either a second network device or encrypted channel ;)
jermar has quit [Ping timeout: 256 seconds]
jgjl has quit [Quit: Textual IRC Client: www.textualapp.com]
AltGr has left #mirage [#mirage]
jermar has joined #mirage
rgrinberg has quit [Ping timeout: 265 seconds]
fgimenez has quit []
w10have has joined #mirage
mort___ has joined #mirage
mort___ has left #mirage [#mirage]
jermar has quit [Ping timeout: 250 seconds]
jermar has joined #mirage
yomimono has joined #mirage
w10have has quit [Ping timeout: 260 seconds]
rgrinberg has joined #mirage
<yomimono> thanks for the merges, haesbaert! D:
<yomimono> er, :D
<yomimono> (probably need to learn to type better before I go making faces at people)
yomimono has quit [Ping timeout: 252 seconds]
agarwal1975 has quit [Quit: agarwal1975]
smondet has quit [Ping timeout: 250 seconds]
yomimono has joined #mirage
agarwal1975 has joined #mirage
yomimono has quit [Ping timeout: 244 seconds]