These are chat archives for arenanet/api-cdi

28th
Apr 2017
Werdes
@werdes
Apr 28 2017 07:08
pi-hole alone is worth owning a pi to be fair.
windwarrior
@windwarrior
Apr 28 2017 12:00
I used to sign mails that were sent by backend systems by -- The Mysterious Sign Painter
or -- The Sign Painter
as a reference to World of Goo, where there are signs that seeminly exist without any clue who put them there. Much the same way as servers magically create email
Nabrok
@Nabrok
Apr 28 2017 13:40
ok, it's pretty cool when you randomly find the bot you wrote out in the wild.
That's kind of terrifying that mysqld crashed.
I've run MySQL in production for almost a decade and never had a hard crash that didn't accompany immediately replacing hardware and restoring from remote backups
Nabrok
@Nabrok
Apr 28 2017 15:16
I can't actually remember any time that mysql has crashed
I'm sure it must have happened
Impending HDD failures will cause it, I guarantee it.
Nabrok
@Nabrok
Apr 28 2017 15:18
The database for my DVR has been running since 2004.
With the nature of my work though, most of my work related databases are only active for a few months at most.
Darrian
@rikkuness
Apr 28 2017 15:20
had some interesting split brain issues with Galera before but never a single daemon crash
Nabrok
@Nabrok
Apr 28 2017 15:23
My DVR stats: Number of shows:
1720
Number of episodes:
35844
First recording:
Monday April 19th, 2004
Last recording:
Friday April 28th, 2017
Total Running Time:
13 years 8 days 9 hrs 56 mins
Total Recorded:
3 years 9 months 29 days 2 hrs 49 mins
Percent of time spent recording:
29%
Of course it has had a few hardware upgrades in that time.
Apart from the (very interesting) technical details implementing this is feasible “if” ANet wants to. I guess they don’t though. Seems like they’re hoping this problem will just go away by itself (Spoiler: It won’t).
mannnnnnn
Nabrok
@Nabrok
Apr 28 2017 15:33
Where's that from?
Nabrok
@Nabrok
Apr 28 2017 15:35
Can't you have the client dump the json to a file?
that's barely better than just OCR'ing the combat log
and it'd still take a couple of weeks to implement
(if it didn't get shot down since it's way outside my normal kingdom)
Nabrok
@Nabrok
Apr 28 2017 15:38
Well, it has the advantage it can be done after the fact, and you don't need to have combat log active in your chat window.
IMO those are barely advantages
David Reeß
@queicherius
Apr 28 2017 15:39
Phss, programming is easy, I wrote hello world once, I can confirm that. An hour tops imo.
Nabrok
@Nabrok
Apr 28 2017 15:39
haha
If I'm gonna do it I'm embeding a TLS cert and adding a websocket implementation
So that you don't need a native app
David Reeß
@queicherius
Apr 28 2017 15:40
^ Yeah, that'd be useful for things like 3rd party chat apps too (which at least I get way more requests for than dps meters)
See, chat is one of those things that I could hypothetically implement as a web service
That would actually only take like week or two to implement I think.
(maybe four)
Darrian
@rikkuness
Apr 28 2017 15:41
mailboxes? :p
like the in-game mailboxes?
Nabrok
@Nabrok
Apr 28 2017 15:42
I remember chat was one of the things mentioned in the extended experience.
Darrian
@rikkuness
Apr 28 2017 15:42
yeah my guild leader is forever having a moan about that, I never use it myself but yunno
I was talking about guild chat
so, mailboxes are a nightmare actually
I can provide an endpoint that dumps out the data but it won't actually update unless you're logged in or I spend an extra two weeks figuring out how geese work
(my understanding is that mail is delivered via geese)
David Reeß
@queicherius
Apr 28 2017 15:42
:D
the benefit is that pending TP coin/items are also delivered via geese, so it'd be a two-for
two-fer?
lye @lye sips coffee
TFW you get into the office and you have a handful of V8 instances each consuming 8GB of RAM
Nabrok
@Nabrok
Apr 28 2017 15:44
Another game I'm playing a lot right now is Elite Dangerous ... their 3rd party apps went on strike yesterday.
oof
Nabrok
@Nabrok
Apr 28 2017 15:44
Most of the major ones shut themselves down to protest lack of communication/feedback channels.
Eearslya Sleiarion
@Eearslya
Apr 28 2017 15:52
they should make a gitter
lye @lye sips coffee
so my new personal stance on further API development is that if it's not WvW-related I don't want to implement it
Eearslya Sleiarion
@Eearslya
Apr 28 2017 15:54
WvW is simpler? XD
Nabrok
@Nabrok
Apr 28 2017 15:54
They had a discord apparently, but the devs didn't use it.
actually crap that doesn't exclude combat stats, would it
Eearslya Sleiarion
@Eearslya
Apr 28 2017 15:54
:D
Is slotted tactics already up as an issue?
Hmm.
I thought slotted tactics were already exposed
Eearslya Sleiarion
@Eearslya
Apr 28 2017 15:55
Oh, are they?
yeah, guild_upgrades
Eearslya Sleiarion
@Eearslya
Apr 28 2017 15:55
Whoopsie.
Darrian
@rikkuness
Apr 28 2017 15:56

so my new personal stance on further API development is that if it's not WvW-related I don't want to implement it

so I take it that mean the WvW PR is in now then? ;)

I wanted to draft the frontend implementation before the PR; I've got it open now
Eearslya Sleiarion
@Eearslya
Apr 28 2017 15:56
Guild K/Ds? :D
Nabrok
@Nabrok
Apr 28 2017 15:56
points_tick corresponds to the tier, doesn't it?
yes, but the ticks/tier can change based on game server configuration
so points_tick is provided in case we ever change it
Darrian
@rikkuness
Apr 28 2017 15:57
ay no pressure :) I do think it'll be a cool feature, at the moment being in a guild that doesn't PTFO is a bit weird, there's no real metric to know if you're "doing well" yunno
Also note that yaks_delivered/tier can vary as well (though not without a game build IIRC) -- the content-driven values are available from https://api.guildwars2.com/v2/wvw/upgrades?ids=all
@Eearslya
    fetchMatchStats(req, state, match, (err, rawStats) => {
        if(err) {
            return cb(err);
        }

        // Aggregate the data for each team (e.g., exclude the team and sum the
        // deaths/kills of the other team values).
        // TODO
    });
Nabrok
@Nabrok
Apr 28 2017 15:58
I don't actually see tier
crap, do the objectives not reference the objective upgrade id
each objective should have an upgrade_id that references /v2/wvw/upgrades that has the yaks_required for each tier
Nabrok
@Nabrok
Apr 28 2017 16:01
ok, /v2/wvw/objectives has that, but not in /v2/wvw/matches
Darrian
@rikkuness
Apr 28 2017 16:01
will the team based K/D be under a guild authenticated endpoint or generic for wvw?
Archomeda
@Archomeda
Apr 28 2017 16:01
urgh, somehow my debian server keeps losing internet access on random intervals
for NO REASON
it started earlier this week
@rikkuness the top N guilds by either kill count or KDR will be public
I haven't decided if querying the stats for a specific guild should require auth
Archomeda
@Archomeda
Apr 28 2017 16:02
i know this is not debian support, but does anyone have tips on how to find the root cause? :D
@Archomeda kind of hard to diagnose without more details, but maybe this will help
Archomeda
@Archomeda
Apr 28 2017 16:03
-_-
I can't help myself sometimes :clap:
Archomeda
@Archomeda
Apr 28 2017 16:04
not gonna install a different distro just for this issue
it's weird, since it does have LAN access
it's a different OS entirely ;_;
Archomeda
@Archomeda
Apr 28 2017 16:04
so this implies a firewall issue
could be an issue with the routing table
Archomeda
@Archomeda
Apr 28 2017 16:05
hmm
timing could line up with DHCP lease expiry
next time it happens grab the output of netstat -rn (or whatever linux uses to dump the routing table)
Archomeda
@Archomeda
Apr 28 2017 16:06
well, i'm currently ssh'd in it while it has no internet access :D
it happened 1.5h ago again... i rebooted the server like 12h ago or something
hmm, AFAIK a routing table issue would break all TCP connections that no longer have a route
Archomeda
@Archomeda
Apr 28 2017 16:07
ehm... what's the normal output of netstat -rn?
like i have only 6 entries
it should have a handful of entries, one of them should be marked DEFAULT
the DEFAULT one should be your router's IP address
windwarrior
@windwarrior
Apr 28 2017 16:07
Phss, programming is easy, I wrote hello world once, I can confirm that. An hour tops imo.
I wrote hell world once
a bug in a single line program
that must be first
Archomeda
@Archomeda
Apr 28 2017 16:08
of which 2 are tun0, 1 docker0, 2 random interface names (probably docker), and only 1 eth0
you don't see a default anywhere in there
Archomeda
@Archomeda
Apr 28 2017 16:08
nop
maybe linux's netstat output is wildly different from freebsd
ugh
do you have an entry with the G flag
Archomeda
@Archomeda
Apr 28 2017 16:09
yes, one, UG
destination should be 0.0.0.0/0 and gateway should be your router's IP
(and iface should be eth0)
Archomeda
@Archomeda
Apr 28 2017 16:09
10.9.8.0 10.9.8.2 255.255.255.0 UG 0 0 0 tun0
definitely not it
uhh
are you using a VPN
Archomeda
@Archomeda
Apr 28 2017 16:10
no
IIRC that's what tun devices are for
Archomeda
@Archomeda
Apr 28 2017 16:10
then wth is this... :<
maybe docker is doing something wonky with the routing table
Archomeda
@Archomeda
Apr 28 2017 16:10
maybe
lemme stop the docker service
Is 10.9.8.0/24 your local-only subnet for docker, or is that your LAN's subnet?
Archomeda
@Archomeda
Apr 28 2017 16:10
i have no clue
is your router within that subnet?
Archomeda
@Archomeda
Apr 28 2017 16:11
my subnet is 192.168.2.0/24 (24 right?)
actually.
windwarrior
@windwarrior
Apr 28 2017 16:11
... on the risk of sounding dumb, no entries in your journal when the internet dies?
@Archomeda that entry is fine since the destination is restricted to the subnet
Archomeda
@Archomeda
Apr 28 2017 16:11
this is the only eth0 is this one: 192.168.2.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
There should be an entry with the destination 0.0.0.0 as a catch-all
Archomeda
@Archomeda
Apr 28 2017 16:11
there is not :(
@windwarrior where can i find that? :D
windwarrior
@windwarrior
Apr 28 2017 16:12
journalctl
if you have debian 8
try sudo route add default gw $ROUTERIP eth0
windwarrior
@windwarrior
Apr 28 2017 16:12
(please dont shoot me, I know its systemd)
(you can undo this with sudo route delete default gw $ROUTEIP eth0)
Archomeda
@Archomeda
Apr 28 2017 16:13
oh god, gitlab errors all over the place since it can't connect to the internet :D
(where $ROUTERIP is your router's IP)
after the route add, you should have a new entry in netstat -rn
Archomeda
@Archomeda
Apr 28 2017 16:16
@lye, well adding that route definitely added the 0.0.0.0 as destination
and pinging stuff works again
okay, so the symptom is that something is clearing the default route
Archomeda
@Archomeda
Apr 28 2017 16:16
apparently
that's normally set by either (1) hardcoded network configuration, or (2) whenever a DHCP lease is renewed
Archomeda
@Archomeda
Apr 28 2017 16:16
i'm currently looking through journalctl
I don't know what journalctl is :(
windwarrior
@windwarrior
Apr 28 2017 16:17
syslog :P
fair
windwarrior
@windwarrior
Apr 28 2017 16:17
but then in systemds own crazy binary format
of course it does
you're better off checking your DHCP server's configuration IMO
since the default route is included in the DHCP lease
Archomeda
@Archomeda
Apr 28 2017 16:18
well, i do have eth0 and wlan0, eth0 is set to static ip, and wlan0 to dhcp; but i toyed around the wlan0 like a month ago and it didn't have any issues
and now it's having issues since a few days ago
uhh, if you're using a static ip assignment you'll also need a static default route
Archomeda
@Archomeda
Apr 28 2017 16:18
let me actually check if i set it static on the server, or my router
since you've got multiple network interfaces, it's way easier to do static assignment on the router
(and let the client think it's being dynamically assigned)
that way the default gateway is always set when you switch between nics
Archomeda
@Archomeda
Apr 28 2017 16:19
yeah, but i'm unsure if this router actually supports it
r-really? that'd be surprising
Archomeda
@Archomeda
Apr 28 2017 16:20
it does support it, but i think it didn't a few years back
cheap isp routers >_>
I swore those off a long time ago, so I can't help you there :<
windwarrior
@windwarrior
Apr 28 2017 16:21
get xs4all, get a fritzbox
(dutch ISP that is not evil^tm)
nah just build yourself a gateway with a microitx/nanoitx board
windwarrior
@windwarrior
Apr 28 2017 16:22
I sometimes like to have devices that I have a hard time breaking
depending on your ISP, buy your own cable modem or whatever else; you'll have to call support and register the MAC against your account but it's no biggie
windwarrior
@windwarrior
Apr 28 2017 16:22
so, a consumer router is great, because it usually doesnt just explode
Archomeda
@Archomeda
Apr 28 2017 16:22
this is my current /etc/network/interfaces
allow-hotplug eth0
iface eth0 inet static
        address 192.168.2.3
        netmask 255.255.255.0
        gateway 192.168.2.1
yeah but my gateway is about the same size but has 8GB of RAM and is running an i3
Archomeda
@Archomeda
Apr 28 2017 16:23
i haven't touched that file in a while
windwarrior
@windwarrior
Apr 28 2017 16:23
and is custom, and configured by you
so, if it breaks, its your fault right :P
yeah, but I also know how to fix it :D
windwarrior
@windwarrior
Apr 28 2017 16:23
my netgear box, if it doesnt work, I nag support :')
Archomeda
@Archomeda
Apr 28 2017 16:23
yes, i added those lines like what... when i installed debian
yeah, that looks fine, hmm.
I'm honestly not sure why you'd lose the default route
Archomeda
@Archomeda
Apr 28 2017 16:24
me neither
unless docker is messing around
might be, but that'd be surprising
actually docker shouldn't surprise me
Archomeda
@Archomeda
Apr 28 2017 16:24
hold on, let me get a part of my log
windwarrior
@windwarrior
Apr 28 2017 16:25
dont get my wrong, I like fiddling with stuff, but some stuff I just like to have a party to nag to... especially for stuff that I need to have high availability from (my router)
Archomeda
@Archomeda
Apr 28 2017 16:25
Apr 28 16:41:31 debianserver dhclient[593]: DHCPREQUEST on wlan0 to 255.255.255.255 port 67
Apr 28 16:41:31 debianserver dhclient[593]: DHCPNAK from 192.168.2.1
Apr 28 16:41:33 debianserver kernel: docker0: port 1(veth24fe5b9) entered disabled state
Apr 28 16:41:34 debianserver kernel: docker0: port 1(veth24fe5b9) entered disabled state
Apr 28 16:41:34 debianserver kernel: device veth24fe5b9 left promiscuous mode
Apr 28 16:41:34 debianserver kernel: docker0: port 1(veth24fe5b9) entered disabled state
Apr 28 16:41:34 debianserver kernel: aufs au_opts_verify:1570:dockerd[31511]: dirperm1 breaks the protection by the permission bits on the lower branch
Apr 28 16:41:34 debianserver kernel: device veth56c85f0 entered promiscuous mode
Apr 28 16:41:34 debianserver kernel: IPv6: ADDRCONF(NETDEV_UP): veth56c85f0: link is not ready
Apr 28 16:41:34 debianserver kernel: docker0: port 1(veth56c85f0) entered forwarding state
Apr 28 16:41:34 debianserver kernel: docker0: port 1(veth56c85f0) entered forwarding state
Apr 28 16:41:34 debianserver kernel: docker0: port 1(veth56c85f0) entered disabled state
Apr 28 16:41:35 debianserver kernel: IPv6: ADDRCONF(NETDEV_CHANGE): veth56c85f0: link becomes ready
Apr 28 16:41:35 debianserver kernel: docker0: port 1(veth56c85f0) entered forwarding state
Apr 28 16:41:35 debianserver kernel: docker0: port 1(veth56c85f0) entered forwarding state
Apr 28 16:41:35 debianserver dockerd[25970]: time="2017-04-28T16:41:35.22678975+02:00" level=error msg="containerd: notify OOM events" error="cgroup path for memory not found"
Apr 28 16:41:36 debianserver ntpd[1535]: Deleting interface #65 veth24fe5b9, fe80::50b9:ffff:fe7c:6136#123, interface stats: received=0, sent=0, dropped=0, active_time=40 secs
Apr 28 16:41:36 debianserver ntpd[1535]: peers refreshed
after that, gitlab runner is complaining about connectivity issues
windwarrior
@windwarrior
Apr 28 2017 16:27
DHCPNAK that doesnt sound good
but a request on wlan0
eh, networking not my speciality haha
Archomeda
@Archomeda
Apr 28 2017 16:28
i just wonder why docker has so many entries
maybe because the connectivity is weird
i dunno
it creates virtual network devices
Archomeda
@Archomeda
Apr 28 2017 16:28
yes, but... Deleting interface #65?
why is it talking over wlan0 though
Archomeda
@Archomeda
Apr 28 2017 16:28
let me just take down wlan0 and see if it keeps working then
how does wlan work on linux? can you just destroy the interface?
ifconfig wlan0 destroy
I'm betting that whole thing works wildly differently from freebsd
Archomeda
@Archomeda
Apr 28 2017 16:29
did ifconfig wlan0 down
that might do the trick; you've just gotta wait until <mysterious event> occurs again
it looks like docker is creating a veth device which is binding to wlan0 instead of eth0
Archomeda
@Archomeda
Apr 28 2017 16:31
and commented out the config in /etc/network/interfaces
a NAK'd DHCP request could cause the routing table to be cleared though
windwarrior
@windwarrior
Apr 28 2017 16:31
(PSA, the authors of linuxes ifconfig state that you should use ip instead)
Archomeda
@Archomeda
Apr 28 2017 16:31
hmm, i don't know my networking
@windwarrior that's linux for you
windwarrior
@windwarrior
Apr 28 2017 16:36
why does dhclient get a NAK on that wlan0 interface tho
Eearslya Sleiarion
@Eearslya
Apr 28 2017 16:40
@Archomeda ip route ?
windwarrior
@windwarrior
Apr 28 2017 16:45
(I really dont get that NAK, your machine gets offered an IP, it says, right I want that one (=DHCP request), and then gets told no... how)
(is your computer being rogue and not requesting any of the IPs it has been offered)
@windwarrior I'm guessing it's requesting a renewal of it's existing lease from eth0
if the wlan and lan are on different subnets (with different dhcp servers), the wlan's dhcp will be like "no i can't renew that lease" with a nak
dhclient should then ask for a new lease but it doesn't seem like it does
or maybe it gets one and the wlan's dhcpd is mis-configured, dunno
windwarrior
@windwarrior
Apr 28 2017 16:52
oh so its sending its renewel of eth0 over wlan0?
could be
might be some other reason, but that's the only reason I can think of that would result in a nak
Archomeda
@Archomeda
Apr 28 2017 16:52
well lol, that's interesting :P
windwarrior
@windwarrior
Apr 28 2017 16:52
yeah exactly
I dont see many reasons for a DHCP server to reject a DHCP request
because it makes the offer first
you could blacklist the MAC, I think
pretty sure that'd yield a NAK
nack? I need more sleep I think, this coffee isn't working
Archomeda
@Archomeda
Apr 28 2017 16:53
oddest thing, the MAC + IP did show up in the list of connected devices in my router
before i took it down
windwarrior
@windwarrior
Apr 28 2017 16:54
NAK :P
Archomeda
@Archomeda
Apr 28 2017 16:54
maybe it was still a leftover i dunno
windwarrior
@windwarrior
Apr 28 2017 16:54
for consistency reasons with TCP
I didn't realize TCP even had a NAK
windwarrior
@windwarrior
Apr 28 2017 16:54
it doesnt
it has ACK
Archomeda
@Archomeda
Apr 28 2017 16:54
i had a course on networking once, i can't remember a thing
windwarrior
@windwarrior
Apr 28 2017 16:54
with a C
wouldn't it be better as NACK then?
windwarrior
@windwarrior
Apr 28 2017 16:55
yeah, but it isnt
thats why it is inconsistent :p
okay people, let's get out the time machine
Archomeda
@Archomeda
Apr 28 2017 16:55
don't these thing only have like 3 characters only?
we've got some heads to knock around
@Archomeda they're not sent as strings over the network :P
Archomeda
@Archomeda
Apr 28 2017 16:55
i know
darthmaim
@darthmaim
Apr 28 2017 16:56
even the new tcp flags all have 3 characters :S
windwarrior
@windwarrior
Apr 28 2017 16:56
yeah, TCP ACK is just a bit in flags
Archomeda
@Archomeda
Apr 28 2017 16:56
but someone made abbreviations of them
and they're all three characters so they lined up nicely in the reference implementation, I'm sure
Archomeda
@Archomeda
Apr 28 2017 16:56
and might have thought, let's have only 3 characters for all flags, because why not
TCP is just SYN/ACK/RST/FIN, right?
windwarrior
@windwarrior
Apr 28 2017 16:56
I suggest to send negative acknowledgements as the angry emoticon
s/emoticon/emojji
I GIVE UP
(mutters something about making typos in typo corrections)
and yeah I think so
oh it's got a lot more than that
darthmaim
@darthmaim
Apr 28 2017 16:57
    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |          Source Port          |       Destination Port        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                        Sequence Number                        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Acknowledgment Number                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Data |           |U|A|P|R|S|F|                               |
   | Offset| Reserved  |R|C|S|S|Y|I|            Window             |
   |       |           |G|K|H|T|N|N|                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |           Checksum            |         Urgent Pointer        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Options                    |    Padding    |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                             data                              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
CWR/ECE/URG/PSH
darthmaim
@darthmaim
Apr 28 2017 16:58
ahh, copied the original rfc
I wonder how often URG/PSH are actually used
darthmaim
@darthmaim
Apr 28 2017 16:58
there are 2(?) extensions by now
Actually I wonder if you can even use them in most userlands
(without raw socket support)
Archomeda
@Archomeda
Apr 28 2017 16:59
well, between the time i had run route and took down wlan0 i'm seeing like DHCPNAK every 10 seconds or so from wlan0 in my logs :P
windwarrior
@windwarrior
Apr 28 2017 16:59
what is urgent supposed to do?
darthmaim
@darthmaim
Apr 28 2017 16:59
exactly what you think it does
mark the package as urgent
@windwarrior I think it's for OOB data
Archomeda
@Archomeda
Apr 28 2017 16:59
Apr 28 18:15:33 debianserver dhclient[593]: DHCPREQUEST on wlan0 to 255.255.255.255 port 67
Apr 28 18:15:33 debianserver dhclient[593]: DHCPNAK from 192.168.2.1
Apr 28 18:15:41 debianserver dhclient[593]: DHCPDISCOVER on wlan0 to 255.255.255.255 port 67 interval 5
Apr 28 18:15:41 debianserver dhclient[593]: DHCPREQUEST on wlan0 to 255.255.255.255 port 67
Apr 28 18:15:41 debianserver dhclient[593]: DHCPOFFER from 192.168.2.1
Apr 28 18:15:41 debianserver dhclient[593]: DHCPNAK from 192.168.2.1
it was like stuck in this loop
It is possible to interrupt or abort the queued stream instead of waiting for the stream to finish. This is done by specifying the data as urgent. This tells the receiving program to process it immediately, along with the rest of the urgent data.
@Archomeda I thought you ifconfig wlan0 down'd?
Archomeda
@Archomeda
Apr 28 2017 17:00
yes, this was before i took it down
did something bring the interface back up?
ah fair
Archomeda
@Archomeda
Apr 28 2017 17:00
and after i added the default route
that'll definitely clear your default route
wait what
I'm not sure why you're getting both an offer and a nak
Archomeda
@Archomeda
Apr 28 2017 17:01
/shrug
Quaggan BooOOoot
@QuagganBooOOoot
Apr 28 2017 17:01
¯\_(ツ)_/¯
Archomeda
@Archomeda
Apr 28 2017 17:03
is there a way to only show dhclient?
i tried journalctl -u dhclient but it's empty :P
windwarrior
@windwarrior
Apr 28 2017 17:03
journalctl -u dhclient iirc
uhh
wait
yeah it should be -u
Archomeda
@Archomeda
Apr 28 2017 17:06
oh my god, dhclient is still trying to DHCPDISCOVER but fails because the network is down, hmm
darthmaim
@darthmaim
Apr 28 2017 17:06
just restart it?
windwarrior
@windwarrior
Apr 28 2017 17:07
and yeah @lye you are right push and urgent
and here I was thinking order was sacred in TCP
Archomeda
@Archomeda
Apr 28 2017 17:07
apparently sudo dhclient -r is a thing
ChieftainAlex
@ChieftainAlex
Apr 28 2017 17:08
there isn't a tp bug where sometimes it returns less items than requested is there?
windwarrior
@windwarrior
Apr 28 2017 17:08
TP usually recovers its mistakes later
like when you're buying/selling stuff on the TP?
It should pop up a yellow "partial success" if it couldn't fully complete the order
Archomeda
@Archomeda
Apr 28 2017 17:09
well @lye, i think i can say that it's very likely that DHCPNAK is the issue
When buying, it might create buy orders for the remaining? I don't remember how it works and I implemented the UI for it >_>
Archomeda
@Archomeda
Apr 28 2017 17:09
the log starts spitting out those things at 16:23
Did you run docker something at 16:22?
Archomeda
@Archomeda
Apr 28 2017 17:10
which is like a few minutes before WAN was inaccessible
ChieftainAlex
@ChieftainAlex
Apr 28 2017 17:10
i meant the tp api, I guess it's a bug in my code that only fires on fast computers.
Archomeda
@Archomeda
Apr 28 2017 17:10
nope
Might be a cache coherency issue if it's on the API
E.g., if /v2/commerce/prices doesn't match /v2/commerce/listings -- they're cached separately
Archomeda
@Archomeda
Apr 28 2017 17:11
16:41 the server lost internet connection
but why that DHCPNAK... i don't understand
maybe the router wants to be thrown out of the window
hmm. For showing the top N guilds by KDR, I'm tempted to use a wilson score or something rather than just the raw KDR value.
Since a bank guild with 1 kill, 0 deaths would have infinite KDR.
And that's not very useful.
windwarrior
@windwarrior
Apr 28 2017 17:14
Calculates the lower bound of the binomial proportion confidence interval as calculated by the Wilson score interval.
right
yeah.
windwarrior
@windwarrior
Apr 28 2017 17:14
the first 4 words are fine
word 5 too
Basically, kills/deaths are a binomial proportion, right?
So there's a confidence interval for the actual KDR given a set of samples (the actual kills/deaths)
As we get more data (more kills/deaths) the confidence in our calculation gets higher
windwarrior
@windwarrior
Apr 28 2017 17:16
right
So the wilson score uses the lower bound of that confidence interval -- for low k/d numbers the confidence interval is very large
So a low k+d will score lower than a high k+d given the same kdr
windwarrior
@windwarrior
Apr 28 2017 17:16
oh right, so if you have a K/D of 1/0 it says, eh I dont have enough data so I am not gonna give you a K/D of infinite
Yeah.
Archomeda
@Archomeda
Apr 28 2017 17:29
anyway, i'll see how long this persists; thanks @lye and @windwarrior for the help :D
windwarrior
@windwarrior
Apr 28 2017 17:30
"helping"
more like rubber ducking :')
Archomeda
@Archomeda
Apr 28 2017 17:30
heh
windwarrior
@windwarrior
Apr 28 2017 17:31
pretty nifty btw @lye. Way more elegant then filtering people with k+d < n for some reasonable n
yeah that was gonna be my original approach but i decided the wilson score would be better
it's what reddit uses (roughly) to sort comments/links
(or used to, at least)
windwarrior
@windwarrior
Apr 28 2017 17:34
the math behind it is still a bit wonky for me (I am .... "great" at statistics) but it indeed sounds like the proper solution ^^
hey man, npm i --save wilson-score
wilson(killTotal, killTotal + deathTotal)
windwarrior
@windwarrior
Apr 28 2017 17:35
hehehe :>
thats not how my analytical mind works
understand first, use later
that took longer for me to find than I expected
darthmaim
@darthmaim
Apr 28 2017 17:37
windwarrior
@windwarrior
Apr 28 2017 17:38
that moment that you really need to go to the supermarket cus you are starving
brb :')
darthmaim
@darthmaim
Apr 28 2017 17:38
.o/
Archomeda
@Archomeda
Apr 28 2017 17:40
hmm @lye, seems like the wvw server relinking is happening today?
i think so yeah
did the API update
oh god
Archomeda
@Archomeda
Apr 28 2017 17:41
ehm, isn't EU reset in like 20 minutes?
i thought it was 40 minutes ago
okay phew was gonna say "doesn't look like the links changed"
Archomeda
@Archomeda
Apr 28 2017 17:41
end_time": "2017-04-28T17:58:00Z"
20 minutes heads-up :D
Darrian
@rikkuness
Apr 28 2017 17:42
aw I missed all the Linux networking fun!
Archomeda
@Archomeda
Apr 28 2017 17:42
@rikkuness it wasn't fun D:
yeah you did, but don't worry, what breaks once will break again ;_;
Darrian
@rikkuness
Apr 28 2017 17:42
I love that shit :p
Nabrok
@Nabrok
Apr 28 2017 17:43
I'm not sure what we'll be doing for wvw tonight.
the same thing we do every friday night, @Nabrok
try and take over the stonemist castle
Darrian
@rikkuness
Apr 28 2017 17:43
try to take over the...
oh
Nabrok
@Nabrok
Apr 28 2017 17:43
We've been waiting to get on JQ, but it hasn't opened
My guild doesn't go to EB very often.
Darrian
@rikkuness
Apr 28 2017 17:44
mine neither, or desert BL ;p
Nabrok
@Nabrok
Apr 28 2017 17:44
We do desert a lot
Darrian
@rikkuness
Apr 28 2017 17:44
oh really? everyone over here seems to hate it and idk why, I've no beef with it
I'm gonna swipe some energy drinks so if our OCX guilds don't show up this week (and we're not against mag) I'll have them on-hand to pugmand all night
Nabrok
@Nabrok
Apr 28 2017 17:45
Yeah, it's our preferred map really.
and if I'm pugmanding it's usually on EBG
Nabrok
@Nabrok
Apr 28 2017 17:45
But maguuma guilds never go there unless there's no other fights.
So often it's just flip everything back and then go to another border.
Darrian
@rikkuness
Apr 28 2017 17:46
did I hear there was a relinking thing happening today?
Nabrok
@Nabrok
Apr 28 2017 17:46
yeah
the borders get really boring by OCX
all t3 keeps and stuff
easier to get good fights in EBG and just port back to home BL to defend when needed
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:48

"helping"
more like rubber ducking :')

one of my coworkers actually got me a duck

Nabrok
@Nabrok
Apr 28 2017 17:48
Well, when you have 20-30 people plus militia starting at the same time, you can't get that many people on EB except for at reset.
Darrian
@rikkuness
Apr 28 2017 17:48
idk if it's an EU thing but we find it impossible to play EBG, the rubber banding is insane
do you have 20-30 people starting in OCX?
Nabrok
@Nabrok
Apr 28 2017 17:48
No
We usually start at 8 eastern.
fair yeah, we have a handful of guilds running then
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:49
@lye API request: Siege placement per-server :D
but after all those raids our OCX guilds have been no-shows for the past couple of weeks
@Eearslya I could do that but nah
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:49
Awwwww
you don't need to know the position and rotation and health of every treb on the map
Nabrok
@Nabrok
Apr 28 2017 17:49
You still on Yak's Bend @lye?
yeah
Darrian
@rikkuness
Apr 28 2017 17:49
French servers OP the shit out of siege oh man
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:50
Oh, no no no no
I just meant numbers
x arrow carts placed since reset
@Eearslya like people would freak out, "YB just placed 99 anet guild acs at their hills"
oh, hmm, open an issue for that
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:50
Hmm. I guess that could be abused too, though.
Just numbers
eh I can fuzz 'em a bit
Darrian
@rikkuness
Apr 28 2017 17:51
locations would be totally abused xD haha kinda gives the game away
would be interesting for raw numbers I think.
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:51
A rise in placed catas..It'd definitely make sneaky-caps harder
If it'd be possible to lag the numbers behind
nah I'd add a random value between -5,+5 before caching the entry
so it'd always be changing
and I'd probably bump the cache time up on that to 5 minutes
lagging might also be possible, if a bit tricksy
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:53
how do I copy a gitter message link..?
oh, got it
right click -> copy link on the timestamp
Nabrok
@Nabrok
Apr 28 2017 17:55
So I guess if we stay linked with JQ we'll do reset as normal, but I'm not sure what the plan is if we get a different link.
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:55
#508
Nabrok
@Nabrok
Apr 28 2017 17:55
Especially if we get linked with TC ... that could be awkward.
Eearslya Sleiarion
@Eearslya
Apr 28 2017 17:56
Ooh, I just realized I'm actually marked as a contributor to api-cdi
darthmaim
@darthmaim
Apr 28 2017 17:57
everyone who has commits to master is a contributor
Nabrok
@Nabrok
Apr 28 2017 17:57
I think that happens automatically if your PR gets merged
darthmaim
@darthmaim
Apr 28 2017 17:57
If you have push access you are either "Collaborator" or "<orgname> Member"
If you open a PR and have no commits in master you are "First time contributor"
looks like the API updated EU
darthmaim
@darthmaim
Apr 28 2017 18:01
\o/
Darrian
@rikkuness
Apr 28 2017 18:02
:D
Archomeda
@Archomeda
Apr 28 2017 18:06
nice
@ChieftainAlex for some reason, https://wiki.guildwars2.com/wiki/Widget:Linked_worlds is displaying no results for EU >_>
Also -- would be slightly better to use /v2/wvw/matches/overview?ids=all there -- still has all the data, just a smaller payload
Edgar Doiron
@Coffee4cr
Apr 28 2017 18:08
why are frameworks such a bitch to work with :worried:
(and now it's displaying data again, weird)
windwarrior
@windwarrior
Apr 28 2017 18:11
heisenbug
there weren't any errors in the console or anything, I wonder what the hiccup was
darthmaim
@darthmaim
Apr 28 2017 18:11
they were just missing from /v2/wvw/matches
it only returned the NA matches
oh were they? derp derp
darthmaim
@darthmaim
Apr 28 2017 18:11
yep
windwarrior
@windwarrior
Apr 28 2017 18:12
if (math.random() < 0.01) return; // They will never find the "bug"
but is transitioning from one linked sets of world atomic?
maybe it was quering some dirty state
nah the matches always disappear from the API for a bit at reset
windwarrior
@windwarrior
Apr 28 2017 18:14
yeah so probably it queried EU before the links were properly set
when you get the message in-game about being redirected out the wvw match is considered "over" by the API backend
well the links get set on the API side when the match starts
so if there's no matches running there are no links
Nabrok
@Nabrok
Apr 28 2017 18:14
Used to be you could see who you were matched against in the API before you could in the game, but there doesn't seem to be so much of a difference now.
we switched up where some of the backend data is stored, so now it has to wait for the map instance servers to be available
or something like that, I can't quite remember the details
windwarrior
@windwarrior
Apr 28 2017 18:15
well it could be: query EU -> links get set -> query NA -> display empty EU -> display populated NA
nah the widget just fetches /v2/wvw/matches?ids=all -- if there's no EU matches it displays no EU links
isn't a way to work around that, I think, except maybe to add a message like "looks like it's reset"
windwarrior
@windwarrior
Apr 28 2017 18:17
yield an error in the transition period
{"error": "no matchups active"}
darthmaim
@darthmaim
Apr 28 2017 18:18
but there are matchups
for the other region
windwarrior
@windwarrior
Apr 28 2017 18:52
the mini karka drops again
https://wiki.guildwars2.com/wiki/Consortium_Chest is available in the gemstore again
obv. fake but oh well
imagine a site that would do it :P
Archomeda
@Archomeda
Apr 28 2017 19:07
oh, https://api.guildwars2.com/v2/commerce is offline too when the tp is down :P
is it? that's kind of surprising
Archomeda
@Archomeda
Apr 28 2017 19:08
or is it just the index page? :P
just the index page lul
Archomeda
@Archomeda
Apr 28 2017 19:08
hahaha
i think it used to show the subendpoints
windwarrior
@windwarrior
Apr 28 2017 19:26
sad thing gitter wasnt bought by Mozilla, otherwise we would have isgitteropensourceyet.com
  /v2/wvw/matches/stats/:id/guilds/:guild_id
  /v2/wvw/matches/stats/:id/teams/:team/top/kdr
  /v2/wvw/matches/stats/:id/teams/:team/top/kills
these routes seem kind of long
/v2/wvw/stats?match=1-1&team=blue&sort=kills hmm
Archomeda
@Archomeda
Apr 28 2017 19:41
i do vote sort as a parameter
:D
yeah but it's not a complete list
so the sort parameter would change the actual results
Eearslya Sleiarion
@Eearslya
Apr 28 2017 19:42
/✌️/🏰/📊/:id/👥/:guild_id
fixed
ty
I should just put that in
Eearslya Sleiarion
@Eearslya
Apr 28 2017 19:43
oh boy
windwarrior
@windwarrior
Apr 28 2017 19:45
I dont think the hiearchical approach is bad
long, but very clear and corresponding to other API endpoints
sort as a parameter is fine tho imo
wait, isnt /v2/wvw/matches/:id/stats/guilds/:guild_id more sensible
nevermind, stats exists already so not worth changing
ChieftainAlex
@ChieftainAlex
Apr 28 2017 19:50
@lye I'll take a look at the wvw linked worlds widget, but please bear in mind i only translated it from the german wiki
darthmaim
@darthmaim
Apr 28 2017 19:51
I kinda want ?ids=all to still work
@ChieftainAlex no worries; the issue was just that the EU matches weren't in the API during reset and I am dumb
ChieftainAlex
@ChieftainAlex
Apr 28 2017 20:01
which bit was empty?
the EU table during reset
ChieftainAlex
@ChieftainAlex
Apr 28 2017 20:02
NA maps appeared in overview but EU did not?
right
The "Linked Worlds (EU)" table was just empty
It might be more intuitive if it displayed "data unavailable during reset" or something
ChieftainAlex
@ChieftainAlex
Apr 28 2017 20:09
not too familiar with the wvw api, NA matches begin with "1" in their id, and EU "2". e.g. "1-1" would be NA?
  • world ids begin with 1 or 2 too
Eearslya Sleiarion
@Eearslya
Apr 28 2017 20:10
windwarrior
@windwarrior
Apr 28 2017 20:15
wait, anets tables are empty
I am scared now
@ChieftainAlex yeah, the match ids are "datacenter-tier"
@Eearslya @Nabrok arenanet/api-cdi#509
windwarrior
@windwarrior
Apr 28 2017 22:14
what is going to be there except top?
average?
i dunno
bottom
maybe I can omit top and just have kills/kdr
darthmaim
@darthmaim
Apr 28 2017 22:17
:+1:
yay thanks :heart:
our QA will be much less confused when double-checking re-links
Eearslya Sleiarion
@Eearslya
Apr 28 2017 22:33
They use the wiki? There's no fancy in-house tool?
XD
Archomeda
@Archomeda
Apr 28 2017 22:56
hahaha
not even using https://api.guildwars2.com/v2/wvw/matches directly pff
/v2/wvw/matches/overview pls is smaller
Archomeda
@Archomeda
Apr 28 2017 22:57
fair :P
ugh, for a personal project I need libpq and libuv working together
and because libuv needs to support IOCP for windows they're totally not gonna play together well
so I'm going to need to run libpq in a separate thread it looks like -_-
let's get rid of computers
Archomeda
@Archomeda
Apr 28 2017 23:04
:wave: @lye it was great knowing you
:wave: likewise, it's been fun
Archomeda
@Archomeda
Apr 28 2017 23:05
when is computer doomsday planned?
windwarrior
@windwarrior
Apr 28 2017 23:05
somewhere in 2032
Eearslya Sleiarion
@Eearslya
Apr 28 2017 23:05
I think you mean 2012*
windwarrior
@windwarrior
Apr 28 2017 23:06
no no, 32, end of unix timestamps
Eearslya Sleiarion
@Eearslya
Apr 28 2017 23:06
Oooh. Good point.
windwarrior
@windwarrior
Apr 28 2017 23:06
eh 2038
On January 19, 2038 03:14:08 GMT
Eearslya Sleiarion
@Eearslya
Apr 28 2017 23:08
What is the plan for that?
windwarrior
@windwarrior
Apr 28 2017 23:08
double timestamps to 64 bits :p?
or do a full IPv6 and make it 128
just in case
oh cool deep impact was actually lost due to overflow of its time counter
link doesnt play nice
the first ariana 5 rocket was also lost due to an integer overflow
Eearslya Sleiarion
@Eearslya
Apr 28 2017 23:14
let's just make the standard size 128 bits and forget it forever
windwarrior
@windwarrior
Apr 28 2017 23:15
or you know, put a slightly less terrible proc in your space ship and do int overflow checks
(they run on pretty old sillicon because small transistors are bad at cosmic rays)
Eearslya Sleiarion
@Eearslya
Apr 28 2017 23:16
just fill it with tube transistors
windwarrior
@windwarrior
Apr 28 2017 23:18
not enough space for that