« first day (1197 days earlier)      last day (2609 days later) » 

12:11 AM
[ SmokeDetector | MS ] Pattern-matching website in answer: Eigen c++ Library: Sparse matrix preallocation by SexyAshley93 on stackoverflow.com
tpu- by Gothdo
tpu- by Undo
CI on 910899e succeeded. Message contains 'autopull', pulling...
 
J F
 
Undo's Raspberry Pi received failover signal.
Restart: API quota is 9995.
 
huh?
!!/location
 
@Undo Undo's Raspberry Pi
New metasmoke user 'JP-Ellis' created
 
@ThomasWard
 
12:25 AM
@angussidney FYI, looks like that change broke something
Not sure what, though
 
!!/rev
 
@Undo that doesn't make sense
 
smokey died didn't it
 
yep
 
12:25 AM
figures
 
@angussidney yeah, that change shouldn't kill anything
 
The regexes in parsing.py don't rely on it or anything
!!/location
 
@angussidney Undo's Raspberry Pi
 
oopsies
 
12:27 AM
!!/failover Aurora
!!/alive
 
@angussidney Of course
 
!!/location
 
@ThomasWard Undo's Raspberry Pi
 
Hmm, stange
strange*
 
I don't think we have a 'force to active' command
it's fine to leave it on Undo's RPi for now
it'll failover to eitheir Aurora or the ESXi
 
12:29 AM
CI on add214d succeeded.
[ SmokeDetector | MS ] Few unique characters in answer, no whitespace in answer, repeating characters in answer: Rotating an image counter clockwise 90 degrees in assembly code by DUDU on stackoverflow.com
 
!!/rev
 
@SmokeDetector that's.... a long way back
 
nobody's pulled :P
!!/pull
 
!!/pull
dammit I got ninja'd
 
12:32 AM
@angussidney No autopull on the standbys yet :P
 
@ThomasWard yeah, intentional
 
[ SmokeDetector ] SmokeDetector started in standby mode at rev add214d (angussidney: Merge pull request #552 from Glorfindel83/master) (running on Undo's Raspberry Pi)
teward/ESXi received failover signal.
 
???????
 
!!/location
 
@angussidney teward/ESXi
Restart: API quota is 9115.
 
12:33 AM
!!/rev
 
wut
 
That's even further back
 
that's brokne
!!/location
 
@ThomasWard teward/ESXi
 
12:34 AM
!!/pyll
 
OK OK we get it smokey you're incredibly robust
 
!!/pull
 
!!/pull
bleh ninja'd
 
I beat you this time :P
 
12:34 AM
@angussidney ESXi isn't as updated as I wish it were :P
 
!!/location
 
i haven't got live syncing working yet - what with locks and crap
 
Why is it that Smokey always goes to standby mode when you !!/pull?
 
teward/Aurora received failover signal.
 
@angussidney not sure, except that it's probably an env. issue
 
12:35 AM
Restart: API quota is 9428.
 
!!/rev
 
@angussidney if we launch with standby mode it'll relaunch in standby mode since it's an arg handed to noparse
 
wonderful we're playing musical standbys
 
well i can fix some of this
 
12:35 AM
Restart: API quota is 9426.
 
OYI AURORA, GET THE HELL OUT OF STANDBY MODE AND BECOME ACTIVE YA PIECE OF ...
there you go
@angussidney Aurora's been launched without 'standby' arguments now
 
!!/tea
 
@quartata brews a cup of jasmine tea for @quartata
 
so things should go better
 
!!/location
 
12:36 AM
@quartata teward/Aurora
 
Note to self: if I ever launch a standby instance, launch it as a normal instance then set it to standby via chat. That way I won't have to play musical standy again
 
I think we'll need a "Get out of standby" exit code
and a "Go into standby" exit code
@angussidney i have an evil nasty 'hack' fix that goes into nocrash.py
it'd work I bet though
if ecode == 3:
    # print "[NoCrash] Pull in new updates."
    git.checkout('deploy')
    git.pull()
    git.submodule('update')
    count = 0
    crashcount = 0
    if "standby" in sys.argv:
        sys.argv.remove("standby")
we'd just have to kick such instances back into standby if they're supposed to be standbys
but that'd not be any different than before
(this isn't pushed yet, I want more than one opinion on it lol)
i really hate the use of argv too... tempted to argparse it...
 
Changes to nocrash have to be manually pulled in, right?
 
but at the same time I'm lazy as shit.
@angussidney yeah
i can do it for Aurora and ESXi very quickly
since both are in SSH windows right now
 
Probably not worth making the change then, since everyone will have to get updated
 
12:45 AM
@angussidney well, if we choose to, I'll drop this into a branch for now :P
 
New metasmoke user 'Ramon Melo' created
 
Well, that works
 
@angussidney I litter the git tree a little with branches, but at least I don't go overboard :P
besides, branches are easy to clean, unlike commit logs :)
 
Git branches cost us nothing, so that's fine
 
oshit i almost deleted the deploy branch o.O
we should probably put deletion protection on it xD
 
12:59 AM
@ThomasWard no, otherwise MS wouldn't be able to automatically merge stuff from master, meaning everything would break down
 
@angussidney uhm
i didn't mean full branch protection, IIRC there's a "Branch Deletion Protection"
or are we saying 'deploy' is regenerated all the time?
 
I'm not entirely sure
 
just curious, 'cause I'm not fluent in the protect functions
 
All I remember is that this is pretty fragile
 
mmm, well at least I didn't break anything oops
 
1:16 AM
CI on 3242881 succeeded.
 
!!/pull
 
[ SmokeDetector ] SmokeDetector started at rev 3242881 (angussidney: Limit reason list to 5 items (+x more)) (running on teward/Aurora)
 
Cool, a numbers-only hash
 
Restart: API quota is 9295.
 
@angussidney checked
 
1:19 AM
@quartata huh?
 
Eh I didn't expect anyone to get the joke
A 4chan tradition when one gets an unusual post number like all 9s
 
@ThomasWard Not sure what problem you're solving here.
 
@Undo we had a set of 'musical standby' chairs earlier because all 3 smokeys had "standby" in their args
ended up with a failover war due to the use of !!/pull
and somehow it not recognizing it right
 
Oh, that's a nocrash issue
 
yep
@Undo but I solved it sorta, with a small fix in the exit code handler for code 3
if "standby" in sys.argv: [remove it from the args] - in its own branch since we don't want to have to restart every smokey in one go
but if you're around i'm happy to merge that in :P
 
1:31 AM
huh, can you do that?
 
well
 
Guess I always thought of argv as immutable
 
@Undo well here's how I usually do things
I usually use argparse, and call [argparserObject].parse_args() which returns a Namespace object
which contains all the args and values
and i store that for referring later on
we can do something similar, at start grab args and store them
then only refer to "runtimeargs" or such
which is usually what I do anyways, rather than relying on sys.argv
 
Yeah, that makes more sense
 
I'd love to use argparse instead of sys.argv but that's a bigger refactor
making it work with a local args var is easy
actually we already half do this
persistent_arguments = list({"charcoal-hq-only"} & set(sys.argv))
we can just use this maybe heh
 
1:38 AM
ew gross
 
I merged Master into it to 'match' the branches up with diffs though.
otherwise it should work fine.
All 3 smokeys will need a restart once it's pulled I'll do it for ESXi and Aurora
your RPi will need a kick though
 
Looks good to me
actually, hold on
 
ooops already merged?
i've not pulled in the code yet
 
Didn't see that persistent_arguments was already a thing
 
@Undo we use it later on
 
1:41 AM
So yeah, should work
 
so i just use that instead
 
CI on d6c6a1d succeeded.
 
pulling on Pi
 
Restart: API quota is 9223.
 
Aurora pulled, ESXi pulled
both restarted
 
oops
i think that broke something
hang on a sec debugging checks
 
Holding off on Pi, then
 
eww
persistent_arguments doesn't work
that's... surprising.
 
Huh?
It... should... work
 
confirmed
hang on doing further testing
% cat test.py
import sys

print "ARGS: %s" % sys.argv
persistent_arguments = list({"charcoal-hq-only"} & set(sys.argv))
print "persistent_arguments: %s" % persistent_arguments
% python test.py standby
ARGS: ['test.py', 'standby']
persistent_arguments: []
 
1:48 AM
wut
 
exactly
 
that's spooky
 
I... think we need to parse this differently
hang on
 
Oh, no, it's working right
Add standby to that first {}
 
ah, yep
 
1:51 AM
& is intersection
 
indeed.
pushed
@Undo I'd still love to move to argparse. It'll change how things are executed, but it'd be a lot 'easier' to weed out bad arguments
but that's a refactor for the future lol
not now.
 
CI on e6adaf5 succeeded.
 
!!/pull
 
Restart: API quota is 9195.
 
there we go
@Undo Good to pull and then run
 
1:53 AM
Sweet, thanks
 
you're welcome, thanks for helping get this pushed to all our instances so quickly :)
 
New metasmoke user 'Kyle Jones' created
 
Oh, there'll need to be explicit exit codes for switching in and out of standby now.
... maybe not
 
@Undo not really
 
Yeah, trying to run through everything in my head.
 
1:58 AM
@Undo the key part is the first if statement
if we have count == 0 which we do for the standby failover, we also have a boolean to determine if we go into standby
which we set to "true" and shortcircuits the first if statement
True or False doesn't get past that True since the 'or' condition is already met
therefore that first if statement works
we can always test by pushing Aurora to standby and forcing something else to active
!!/standby teward/Aurora
 
teward/Aurora is switching to standby
 
Failover for teward/ESXi will be forced on the next ping; probably within 60 seconds.
thank you MS
 
Undo's Raspberry Pi received failover signal.
 
That'd work
 
oops
 
2:00 AM
meh. Let 'em both come up
 
teward/ESXi received failover signal.
 
So now they're both active
!!/location
 
@Undo teward/ESXi
@Undo Undo's Raspberry Pi
Restart: API quota is 9114.
 
!!/standby Undo
 
2:01 AM
looks like it worked :p
 
yay
 
Hrmph
... huh
 
!!/alive
 
@ThomasWard Of course
 
2:02 AM
!!/location
 
@ThomasWard teward/ESXi
 
Yeah, okay then. That'll work.
Might take a stab at simplifying it sometime, but if it works it works
 
@Undo the idea i had was that since it replied to !!/pull we should consider it as 'active' already
 
(specifically, not sure we need switch_to_standby anymore)
@ThomasWard Oh, that's what I wanted to test. Just a sec.
Okay, my Pi comes up as standby....
(jeopardy music)
 
feel free to junk commit and use ESXi
since it's already a standby :p
 
2:03 AM
 
it should then pull and restart in Active then
 
Start a failover...
 
@Undo
elif ecode == 7:
    # print "[NoCrash] Go to Standby Restart Called"
    switch_to_standby = True
This is why we still need switch_to_standby
 
Yay, I got the 4,500th commit to SmokeDetector
 
@ThomasWard Why can't we just add it to persistent_arguments?
 
2:04 AM
Undo's Raspberry Pi received failover signal.
 
Failover signal...
 
@Undo we could actually, since that's just a list.
hmm, not a bad idea.
 
Now, this should break some stuff:
 
Restart: API quota is 9953.
 
!!/reboot
 
Yep
 
oh jeez
 
blurgh
 
!!/location
 
@Undo that's fairly easy to fix though
 
2:05 AM
My Pi hit the invalid literal thing
 
@quartata oyi we know what's going on
 
we're intentionally breaking things to fix :P
 
teward/Aurora received failover signal.
 
yay for Aurora xD
 
2:06 AM
Restart: API quota is 9175.
 
@Undo it shouldn't be too hard to create a new exit code for such !!/reboot cases...
if ("standby" in sys.argv) or w/e to handle which exit code we use
 
@ThomasWard Proposal: Remove switch_to_standby and just add directly to that list. Then, remove it from the list right after github.com/Charcoal-SE/SmokeDetector/blob/master/nocrash.py#L46 and :54
 
@Undo proposal accepted, if it works
 
Thought being that whenever we start in standby mode and that code exits, we're no longer in standby because it never exits in standby mode
I'll do it later then, if you don't beat me to it. Eating dinner in a few.
 
i can do it now
 
2:12 AM
CI on d678491 succeeded.
 
oops?
where's Aurora...
 
!!/lick
 
ooops?
hang on
i broke it lol
fix pushed in
 
can we agree never to play with standby mode ever again
 
teward/ESXi received failover signal.
 
2:16 AM
@quartata Let's agree to disagree
we have to fix the musical standby chairs issue after all
 
Restart: API quota is 9103.
 
Aurora's starting up now
 
Restart: API quota is 9154.
 
@quartata No reason not to - we have Git, and we detect pretty much nothing in this timeframe anyway
 
@Undo you'll need to pull your RPi
 
2:18 AM
also, I still stand by this:
 
and double checking that everything is working wouldn't hurt either right now :P
 
Mar 18 '16 at 19:33, by Undo
observation: projects stop growing when they become change-averse.
2
@ThomasWard done
 
hey lookit all works :)
!!/coffee
 
@ThomasWard brews coffee for @ThomasWard
 
2:23 AM
!!/location
 
@ThomasWard teward/Aurora
 
@Undo and somehow we didn't blow everything up xD
 
3:07 AM
@ArtOfCode Where can I find docs on what SE chat sends through the websocket?
 
3:18 AM
@quartata I don't think those exist anywhere, unfortunately.
 
how did you people make CE then
 
Reverse engineered it, but I don't think we documented it anywhere
by 'we' I mean 'people who aren't me'
 
Oh well. CE's source isn't too horrid.
 
Mar 18 '16 at 19:30, by ArtOfCode
We should submit Smokey to CR and see what they make of it.
We should actually drop into their chat and find out what they think of that idea :P
 
@angussidney I bet they wouldn't like doing the whole project at once, but I've considered throwing smaller bits to them.
bodyfetcher, for example
 
3:25 AM
Looks like it's identical to what /events gives but you can't send messages :(
 
@Undo you should let me post it - I am always over there xD
throws sticks at Code Review just for no apparent reason
 
@Undo and plus, I don't think that our entire codebase will make it inside the 30,000 char limit
 
@angussidney they're special case with a higher limit nearer 60k
give or take 5k
but you're not wrong
 
Can someone on unix run a character count on the smokey repo?
 
anyone want to give me a one liner to use :P
 
3:30 AM
    9756 LICENSE-APACHE
    1056 LICENSE-MIT
    2225 README.md
     333 RevertedMode.md
     411 StandbyMode.md
    1546 apiCalls.txt
    2650 apigetpost.py
    2682 apigetpost.pyc
       6 autoIgnoredPosts.txt
    5087 bad_keywords.txt
      39 blacklistedUsers.txt
     257 blacklisted_usernames.txt
   15371 blacklisted_websites.txt
   16075 bodyfetcher.py
   10148 bodyfetcher.pyc
    1967 bodyfetcherQueue.txt
   62703 chatcommands.py
   51563 chatcommands.pyc
   10766 chatcommunicate.py
    7721 chatcommunicate.pyc
from wc -c
 
yeah we'll blow it up
but
 
*.py is 247580 total
 
@Undo exclude .pyc and .p
 
so a little more than 30k :P
 
@Undo and make sure you run the .txt-to-.p conversion script lol
 
3:31 AM
@Undo y'know, just slightly
 
 
1 hour later…
4:57 AM
!!/lick
 
@angussidney licks ice cream cone
 
!!/tea
 
@Jamal brews a cup of chamomile tea for @Jamal
 
^_^
 

« first day (1197 days earlier)      last day (2609 days later) »