« first day (1638 days earlier)      last day (2187 days later) » 

12:38 AM
[ SmokeDetector | MS ] Offensive answer detected, toxic answer detected: I can not access second monitor anymore by user829223 on askubuntu.com
 
12:54 AM
tpu- by ByteCommander
[ SmokeDetector | MS ] Bad keyword in body: Symfony login with database by Pantherax on stackoverflow.com
 
sd why
 
@thesecretmaster [:44572943] Body - Position 883-887: dbal, Position 901-905: DBAL
fp- by thesecretmaster
 
1:19 AM
@thesecretmaster POST to /api/graphql is returning 422
 
Uh oh
No?
I just tried it.
What query?
 
Did you try it from your browser console?
Logs say 422 Unprocessable, authenticity token failed verification
 
Are you sure you're using the right request format?
 
I, [2018-05-13T01:11:58.330934 #17129]  INFO -- : Started POST "/api/graphql" for MY_IP at 2018-05-13 01:11:58 +0000
I, [2018-05-13T01:11:58.331746 #17129]  INFO -- : Processing by GraphqlController#execute as text/html;application/xml;application/json
W, [2018-05-13T01:11:58.332226 #17129]  WARN -- : Can't verify CSRF token authenticity.
I, [2018-05-13T01:11:58.332523 #17129]  INFO -- : Completed 422 Unprocessable Entity in 1ms (ActiveRecord: 0.0ms)
F, [2018-05-13T01:11:58.333556 #17129] FATAL -- :
Trying from your browser console would get around that, because it automatically includes the credentials
 
@ArtOfCode CSRF?
 
1:23 AM
unless you use fetch, which I'm guessing you didn't
Aye
'cuz I don't send a CSRF token because this is an app, which shouldn't need it
should just need you to skip token verification on that route, @thesecretmaster
 
oh yeah definitely need to skip csrf
 
I just did it from my console
 
Browser? Or CLI?
 
Browser
 
with jQuery?
 
1:25 AM
I'll try from CLI now though
Yeah, with jQuery. Why?
 
2 mins ago, by ArtOfCode
Trying from your browser console would get around that, because it automatically includes the credentials
2 mins ago, by ArtOfCode
unless you use fetch, which I'm guessing you didn't
Your existing cookies jump you through the CSRF check
 
I don't have cookies from code
 
skip_before_action :verify_authenticity_token right
 
aye
 
1:28 AM
@ArtOfCode ....CSRF isn't cookie-based, is it? I thought the token is included in the HTML
(wouldn't cookies defeat the purpose of CSRF protection?)
 
@NobodyNada how so?
yeah, probably
shrug I dunno how Rails does it, honestly
 
@ArtOfCode You want to do it, or should I?
 
go for it
I don't have a Rails env open at the moment
 
@ArtOfCode IIRC CSRF is where an attacker gets your browser to POST to MS from e.g. JS on their site, it makes the request with your cookies and they can do whatever they want on your account
 
@NobodyNada but they can't do that, because CORS
 
1:32 AM
[ metasmoke ] ci/circleci failure on 7ebde0d: Your tests failed on CircleCI
[ metasmoke ] ci/circleci failure on 7ebde0d: Your tests failed on CircleCI
 
@ArtOfCode Wikipedia says HTML forms are exempted from CORS though
 
@NobodyNada shrug
this is probably why Rails doesn't do it with cookies... or maybe half-cookies... I dunno
Rails internals are mostly black magic
 
@ArtOfCode yeah, Rails must do some of it's crazy weird JS trickery
 
probably not even JS
suspect they just know the specs better than I do
 
@ArtOfCode 2 PRs waiting for ya
 
1:36 AM
 
If I remember correctly they use a special parameter in HTML forms and there's something in ujs that takes care of everything else -- this must be it
But shrug
 
By all appearances it should legitimately be caching each individual reason
 
@quartata something?
 
[ metasmoke ] ci/circleci success on 2993c0d: Your tests passed on CircleCI!
 
'tis, ain't it?
 
1:37 AM
What's the default expires_in for that helper
 
[ metasmoke ] ci/circleci success on 2993c0d: Your tests passed on CircleCI!
 
I just tossed in a PR to cache the whole query.
I probably should've looked at the view first though...
 
So if we're already caching that then we should set it to never expire
How do you bust the cache from the controller then though
For when that reason gets a new post
(after_create can go on the posts_reason association? Dunno)
 
[ metasmoke ] ci/circleci success on 734f736: Your tests passed on CircleCI!
[ metasmoke ] ci/circleci success on 734f736: Your tests passed on CircleCI!
 
@quartata never
pretty sure we bust those caches on post create
 
1:40 AM
Ahhh
That's an easy enough fix then
Let me take a peek
 
[ metasmoke-deploy ] deploy started by ArtOfCode
 
Actually shouldn't it be busted on feedback?
Wait I remember something about this
 
@ArtOfCode My next commit to the /reaons PR was to remove the old cache -- it makes the new cache useless.
So... I guess a second PR?
 
Wait leave it all alone
 
Imma let you two sort this one out :P
 
1:41 AM
[ metasmoke ] ci/circleci success on 52e236b: Your tests passed on CircleCI!
 
We need to figure out what the heck is going on here
 
Too late
@ArtOfCode Didn't you just deploy?
 
yup
 
what you merged that
 
yes
 
1:42 AM
...
 
..?
 
Now /reasons results are going to be stale
 
[ metasmoke ] ci/circleci success on 7bdc5b0: Your tests passed on CircleCI!
 
and you probably broke the part where it busted the old cache
 
Yeah, by a couple hours, but it'll be fast.
Also, those number just go to the tiny bar graphs, no?
So stale-ness doesn't really master?
 
1:44 AM
Hah, now I'm getting a HttpStatusCode.NotAcceptable response to graphql. @thesecretmaster what format do I send requests in?
 
It also shows number of posts
 
{query: "ur query", variables: "ur vars"}
 
I think that's for the /reason/id breakdown
 
See? /reasons is so fast now
And what harm is done if the numbers are 3 hours behind?
 
@thesecretmaster JSON, then?
 
1:46 AM
Yes
You can omit vars if your not using them
 
something is super wrong here
 
It's fast, I'm happy.
 
because it should already have been expiring the fragments for /reasons and it wasn't
 
Am I what's wrong?
 
@ArtOfCode do you know where reason.post_count is updated?
@thesecretmaster It should have been just as fast while keeping results fresh if it worked the way it was written. That's why I'm confused
 
1:52 AM
@quartata When it's asked for. In the reaons model.
 
The cache view helper puts in the reason object, which I assume is the key for the cache. Does expire_fragment expire all the fragments associated with the model?
 
@quartata No, because in that case the caches were getting kicked every request, from what I can tell.
 
@thesecretmaster Okay, this is a bug this time: got a 406 to a request formatted precisely like that
 
What's the query?
 
posts(last: 10) {
                autoflagged
                title
                site {
                    site_name
                }
            }
 
1:54 AM
@thesecretmaster caches weren't expired until a feedback was created or destroyed if what Art said about the expiry time is true
do you see a place where the reason cache is invalidated that I'm missing
 
...which, given that it came from my memory of the last time I touched that code, may not be entirely accurate
 
I take it it's a custom helper?
 
what is?
 
@ArtOfCode status-norepro
 
@thesecretmaster status-found
missing outer braces
 
1:55 AM
Ah
 
...or not
 
@ArtOfCode cache
 
no, Rails method
 
Also, turns out it's best practice to use query { the actual query }
@ArtOfCode curl https://metasmoke.erwaysoftware.com/api/graphql --request POST --header "Content-Type: application/json" --data '{"query": "{posts(last:10){autoflagged title site {site_name }}}"}'
Works on my machine
 
likewise
 
1:57 AM
Ohhh wait a second
 
but not from C# for some reason
 
I see the problem
expire_fragment is specific to the view?
OK I think it's being used wrong here...
It says expire_fragment is supposed to take a string but a model is getting passed in
 
Sooo.... I've been running a python script to go from tsv to csv for about an hour now, and no visible progress.
@quartata Post.new.to_s
Ah, but that includes the object ID not the AR id
That'd do it.
 
yeah plus isn't it supposed to be like "views/.../" or whatever
 
That's a choice
 
2:02 AM
Is there any way to see the current Rails cache?
@thesecretmaster I mean this doesn't specify a key so it's going to be whatever one Rails made
 
Yes it does. Rails will try to coerce it to a string
But wait.
You're missing something, I think.
Even if the view caching is done perfectly, the query still happens in the action.
Which is what takes the time.
 
...right. Since it fetches them all at oncr
 
ok
 
2:05 AM
So you gotta stick the cache there.
 
tpu- by thesecretmaster
 
or gather which reasons have gone bad and query for those
 
That would probably take more time, for not much gain.
 
!!/watch-keyword buffalo\-airporttaxi\.com
 
@JoErNanO You don't have code privileges, but I've created PR#2163 for you.
 
2:07 AM
I suppose I'd be yelled at if I did one query per reason, so I do have to do an IN unfortunately
 
!!/watch-keyword buffaloexpresstaxi\.com
 
@JoErNanO You don't have code privileges, but I've created PR#2164 for you.
 
Some reasons are so rare though that they would probably never need to be fetched again
ah here we go
 
Merged SmokeDetector #2163.
 
You'd have to pull all the reasons, check which ones are expired, but you already have all the reasons.
 
2:09 AM
CI on fb4116c succeeded. Message contains 'autopull', pulling...
Restart: API quota is 14974.
Merged SmokeDetector #2164.
Restart: API quota is 14965.
 
@thesecretmaster nope, Reasons.where.not(id: [...cached fragments...]).<rest of that query>
 
[ metasmoke ] ci/circleci success on 252b0fc: Your tests passed on CircleCI!
[ metasmoke ] ci/circleci success on 252b0fc: Your tests passed on CircleCI!
 
Bah, the problem with HTTPS is that I can't inspect it
 
Ehm... If you want to do it...
 
@ArtOfCode ever used mitmproxy?
 
2:11 AM
The trouble is managing the caching, since Rails makes the fragments pretty opaque.....
 
@NobodyNada nope
 
@ArtOfCode Basically you just run it in a terminal and configure your proxy settings to go through localhost:8080
If you visit mitm.it you can install an SSL certificate that lets it inspect HTTPS stuff
 
but HSTS is a thing
shrug might as well try
 
24 secs ago, by NobodyNada
If you visit http://mitm.it you can install an SSL certificate that lets it inspect HTTPS stuff
 
2:59 AM
[ SmokeDetector | MS ] Bad keyword in answer, blacklisted website in answer, link at end of answer: Importing my Apple Music library to Spotify by Kevin Ronk on superuser.com
 
@SmokeDetector tpu- has many previous tp's (from same + other usernames)
 
tpu- by Makyen
 
@SmokeDetector While the question does ask for recommendations, it also asks for quite a bit more than just a recommendation. This answer doesn't address the rest of the question. This is the third post detected in the last 24 hours from this user, or at least a user with this name, advertising for this site.
 
@ArtOfCode domain not detected there?
 
3:18 AM
is SO chat acting up for anyone else?
 
@Machavity The only thing I've noticed was that the most recent SD report did not show in this room until I refreshed the page. It had shown up in another room at least a couple/few minutes prior to my wanting to reply to it here. Both posts now show the same timestamp.
 
@Machavity yeah.
Some runtime error.
 
@Makyen Might not be showing if you're still in the room. Attempting to join one gives a 500 error
Ugh, might be spreading to SE chat now. Getting timeout notices
 
OK. Now I'm seeing something (unable to connect to server) in all rooms in which I already have a tab open.
 
Well, it's bedtime. CVQ can wait. Can't even report to Natty like this
 
3:28 AM
And a "Runtime Error" when trying to enter a new room.
 
@Makyen I'm receiving a runtime error everywhere.
 
Henders/EC2 received failover signal.
 
!!/location
 
NobodyNada/Raspberry Pi received failover signal.
 
!!/location
 
3:40 AM
!!/alive
 
Ooh that's bad. It'll trigger on every failover since it tries to join rooms after then
 
All instances seem to be dead on MS?
 
It doesn't need to join rooms to send messages, hence the failover warning
 
All instances show multiple minutes since last ping.
 
@AshishAhuja yeah they'll all die eventually because it's at the top level where join_command_rooms is called
It's not supposed to, like, error ever
 
3:42 AM
!!/alive
 
@quartata uh, so all instances are now going to be dead?
 
@quartata what happened? anything I can do to fix it?
 
Bots in SEBotics aren't dead, even though they should also connect to the SO chat room.
@NobodyNada something seems to be wrong with SO chat.
 
I'll launch an instance that runs only in here
 
@AshishAhuja The SEBotics ones are separate instances if I'm not mistaken
 
3:43 AM
Fire runs on SO
 
@AshishAhuja MS will failover and cause them to die, one by one
 
It has pretty good error handling once it's running, but it might die if it fails during startup
 
@NobodyNada put commands: false for every room
 
@NobodyNada Natty and Housekeeping seem to be alive in SEBotics.
 
Then it won't join them
But still report there
 
3:44 AM
You can still take commands from here.
 
@NobodyNada don't you run a script to restart FA on crashes?
@quartata they are? lemme check...
 
Yeah OK just turn them off for the SO rooms
 
@AshishAhuja Yes, but if it's already running it's unlikely to crash
 
chat.meta.stackexchange.com appears to be fine. It's only chat.SO
 
(although now that I think about it it will crash if it gets too many errors in a short time)
@quartata On it...
 
3:45 AM
a proud member of the Stackexchange community
 
@skull we can't delete users; I guess you'll have to contact SE.
 
@skull That's nice, but we didn't need it oneboxed here.
 
@quartata my Git is messed up; one moment
 
@skull If you have a problem with the user's name (which you reasonably should), then custom mod-flag one of their posts. If they have no posts, then custom mod-flag one of your own. If you have none on that site, then pick a random post and custom mod-flag. In the flag, explain the problem, in detail, with a link to the user.
 
3:49 AM
@quartata done; Smokey should be on its way up
[20:50:09] ValueError: invalid literal for int() with base 10: 'login?returnurl=http%3a%2f%2fchat.stackexchange.com%2fchats%2fjoin%2ffavorite'
 
Try again
 
@quartata Got the error several times in a row
 
Check your credentials
 
@NobodyNada that happens with CE; I've seen that before.
 
@quartata Should be fine, unless somebody changed them
 
3:55 AM
Keep trying
 
4:39 AM
[ SmokeDetector ] SmokeDetector started in reverted mode at rev c0cee72 (tripleee: MP3 download: add Turkish --autopull) (running on teward/Lunar Eclipse)
Restart: API quota is 19995.
 
Restart: API quota is 14674.
Restart: API quota is 19998.
Restart: API quota is 19993.
Restart: API quota is 19997.
 
Seems to be back online
!!/rev
 
!!/pull
 
4:43 AM
Yeah, what happened?
 
@skull Chat was down.
 
tpu- by Makyen
Restart: API quota is 19987.
Restart: API quota is 19987.
Restart: API quota is 14662.
 
!!/standby teward
 
teward/Lunar Eclipse is switching to standby
teward/Solar Flare is switching to standby
Restart: API quota is 19980.
 
4:43 AM
!!/standby henders
 
Henders/EC2 is switching to standby (@quartata)
Restart: API quota is 19978.
 
!!/location
 
Thanks @Machavity
 
!!/standby teward
 
!!/location
 
@WELZ ArtOfCode/EC2
 
Thanks @Makyen
 
!!/alive
 
[ SmokeDetector | MS ] Bad keyword in answer: encoding XDCAM MXF with FFMPEG by beigubfd on stackoverflow.com
@WELZ Of course
 
4:46 AM
@NobodyNada you can re-enable chatcommunicate
 
And we're back
Ohhhhh
It was @WELZ the whole time
You're the reason I couldn't get Supernova to standby
 
@quartata What?
 
[ SmokeDetector | MS ] Bad keyword with email in answer: Use image file as camera by Sáng on android.SE
fp- by NobodyNada
 
I didn't know chat.SE was back up so I was trying to do a bunch of crap on chat.meta.SE
 
naa- by PeterJ
 
4:49 AM
I ran standby-except Lunar right before you ran standby reward, so everyone went into standby and Supernova failed over and it hurt my brain
 
naa by NobodyNada
 
meh, I didn't even bother checking if chat.meta.se was online -_-
Could've vented there!
 
It was online the whole time
!!/pull
Where were you
 
We have a command !!/standby-except ... I didn't even know - that's useful!
 
4:52 AM
[ SmokeDetector | MS ] URL-only title, bad keyword in body, bad keyword in title, blacklisted website in body, blacklisted website in title, +1 more: mumybear.com/turmeric-forskolin/ by user829270 on askubuntu.com
tpu- by NobodyNada
 
Now that we're back, I'll restore everything to normal on my instance
 
So I just realized something
None of the instances ever died
But they send status pings after logging into chat, which fails
 
That makes sense.
 
@quartata or nocrash kept reviving them
 
So they were crashing and reverting and crashing and reverting
Now that means instances are going to be quite a lot of revs behind
So pull next time you failover
 
4:54 AM
@quartata I pulled them all before running standby.
 
Cool
Yeah you can every one was on a different rev
Race to the initial commit, as it were
 
Why don't we make it autopull when it comes alive?
When it's failover(ed)
 
@WELZ We've discussed that a few times -- TL;DR is that if a commit breaks Smokey, then all our instances die
 
@NobodyNada That sounds fun LIKE HELL
 
5:05 AM
@quartata RPi should be back to normal configuration
 
[ SmokeDetector | MS ] Bad NS for domain in body: WordPress + Woocommerce site crashes when checking out by HardwoodFun on stackoverflow.com
fp- by NobodyNada
 
Good. I think all the instances came back
Yep we're good. Angus's EC2 was down before this
(@angussidney should probably get that back up)
 
@Mak known spam?
 
5:21 AM
@NobodyNada All but 1 hit in MS are spam and all have very similar wording to this post. The question doesn't ask for recommendations. The question asks what settings to use for FFMPEG. The account was created today, apparently just to post that answer. I haven't raised a spam flag, yet. I'm debating between a spam flag and a custom flag, as it doesn't look as much like spam w/o a link to the product.
 
@Makyen Ooh, very good points -- I'd leave that as a comment on MS if I were you BTW
 
5:37 AM
@NobodyNada Good point. I'll leave a comment.
 
5:56 AM
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, pattern-matching website in body, pattern-matching website in title, repeated URL at end of long post: www.garcinialossweight.com/keto-6x by ripsdvc on apple.SE
[ SmokeDetector | MS ] Blacklisted website in body, link at end of body, pattern-matching website in body: This enables in ensuring a by adncedliver on askubuntu.com
tpu- by PeterJ
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, bad pattern in URL body, blacklisted website in body, link at end of body, +1 more: mini authentic alliance has evermore by Foraw aimis on drupal.SE
tpu- by Makyen
[ SmokeDetector | MS ] Bad keyword in link text in body: Shortcut For Adding Bullet Point In Mac's Notes Application by Aaron Miller on apple.SE
fp- by PeterJ
 
6:16 AM
@tripleee @Mithrandir @iBug Just FYI: I've submitted a PR to tighten the blacklist for D\W?Bal to D\W+Bal. d\W?bal was added previously and removed (issue #1757, PR #1763) due to too many FP. Changing to requiring the non-word character still detects all the TP, while not detecting all of the FP.
 
[ SmokeDetector | MS ] URL-only title, bad NS for domain in body, bad NS for domain in title, bad keyword in body, bad keyword in title, +7 more: luxmuscle.com/summit-keto-reviews/ by Abse1992 on askubuntu.com
tpu- by PeterJ
[ SmokeDetector | MS ] Potentially bad keyword in answer: Get web methods dynamically for an asmx service by Tejas Mehta on stackoverflow.com
fp- by Glorfindel
 
7:04 AM
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body, link at end of body: Summer Keto While some put on weight by stenon12 on askubuntu.com
tpu- by PeterJ
 
7:50 AM
[ SmokeDetector | MS ] URL-only title, blacklisted website in body, blacklisted website in title, link at end of body: mumybear.com/summer-keto/ by btrreya on askubuntu.com
tpu- by PeterJ
[ SmokeDetector | MS ] Pattern-matching website in answer: Hosting two domains using only one VPS? by user57718 on stackoverflow.com
naa- by rene
 

« first day (1638 days earlier)      last day (2187 days later) »