Attack of the Cooties
-
The cooties are strong today....
-
weakass attempt to zoom enhance the chart to remove everything but the overall line to get a less crowded view....
-
The sequel to The Phantom Likes and the prequel to Revenge of the Jeff?
-
with a little bit of
Attack on Titan
thrown in, yes.
-
weakass attempt to zoom enhance the chart to remove everything but the overall line
Why not add a checkbox that tells your drawing software to only draw that line?
-
Why not add a checkbox that tells your drawing software to only draw that line?
Slated for version 0.5.2
-
-
There's a post of mine in t/1000 that seems to be the root of a deadlock; no idea why though
-
There's a post of mine in t/1000 that seems to be the root of a deadlock; no idea why though
Probably too much talk about badges.
-
Joking about the Hokey Cokey actually
-
Another entry in the logs:
ActiveRecord::StatementInvalid (PG::TRDeadlockDetected: ERROR: deadlock detected DETAIL: Process 8147 waits for ShareLock on transaction 86319669; blocked by process 5435. Process 5435 waits for ShareLock on transaction 86319670; blocked by process 8147. HINT: See server log for query details. : UPDATE "topics" SET like_count = 1428336 WHERE ("topics"."deleted_at" IS NULL) AND "topics"."id" = 1000) HTTP_HOST: what.thedailywtf.com REQUEST_URI: /post_actions REQUEST_METHOD: POST HTTP_USER_AGENT: SockBot/0.16 (Hazardous Hera; rv:09ced6f; owner:@RaceProUK) @RaceProUK/20150606 username: RaceProUK params: id: 452808 post_action_type_id: 2 flag_topic: false message:
Different IPs than the last time (not included in either post, BTW, for anonymization). But this one matches yours. No clue what all that means, TBH. @sam ?
-
Oh! No, this is cyberparts, I guess. The bots are killing the site!
-
And this is the first one:
ActiveRecord::StatementInvalid (PG::TRDeadlockDetected: ERROR: deadlock detected DETAIL: Process 7947 waits for ShareLock on transaction 86239875; blocked by process 12675. Process 12675 waits for ShareLock on transaction 86239874; blocked by process 7947. HINT: See server log for query details. : UPDATE "posts" SET reply_count = reply_count + 1 WHERE ("posts"."deleted_at" IS NULL) AND "posts"."id" = 452671) TTP_HOST: what.thedailywtf.com REQUEST_URI: /posts REQUEST_METHOD: POST HTTP_USER_AGENT: Mozilla/5.0 (Windows NT 6.3; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/43.0.2357.81 Safari/537.36 HTTP_ACCEPT: application/json, text/javascript, */*; q=0.01 HTTP_REFERER: http://what.thedailywtf.com/t/like-nice-bload-like/1000/52143 username: RaceProUK params: raw: Are you doing the Hokey Cokey? category: 13 topic_id: 1000 is_warning: false archetype: regular reply_to_post_number: 52144 nested_post: true
-
My original logs post:
http://what.thedailywtf.com/t/like-nice-bload-like/1000/52191?u=boomzilla
-
The 'reply count' one seems to be two people posting at the same time (interestingly, I got a 'Sorry, error' message when I first attempted that post); the second must be simultaneous likes.
Neither of which should be triggering deadlocks.
-
Oh! No, this is cyberparts, I guess. The bots are killing the site!
HTTP_USER_AGENT: SockBot/0.16 (Hazardous Hera; rv:09ced6f; owner:@RaceProUK) @RaceProUK/20150606
yep. that's why sockbot sets such a recognizeable user agent.
The bots are killing the site!
... not intentionally.
-
@boomzilla said:
The bots are killing the site!
... not intentionally.
I'm pretty sure he's taking the piss ;)But again, this shows that they're useful; they're exposing issues in Dischorse that would otherwise be missed.
-
I'm pretty sure he's taking the piss
I hope so, but I could be accidentally right.
:this_is_discourse.flv:
-
-
The odds of it being a Discourse fault are significantly higher than it being a bot fault.
-
Agreed, PG deadlocks sound serious.
-
this is cyberparts
No, this is shitty coding with no regard for concurrency.
My best bet, at least.
-
Well, yes, obviously. But I think the way the cyberparts are implemented we have the same user doing stuff from different IPs, which might be an important detail in figuring out the actual problem.
-
A valid point, though if it is true, then it's yet another flaw in the design of Dicsource; it should be IP-agnostic
-
I think it is probably IP agnostic, but my point is that they may be concurrency safe between different users, but not against a single user with multiple concurrent sessions. That's just a WAG, but I've debugged enough things that weren't easy to reproduce to know that little details like that could be important.
-
But I think the way the cyberparts are implemented we have the same user doing stuff from different IPs
correct
they may be concurrency safe between different users, but not against a single user with multiple concurrent sessions
hmm that's a thought....I've debugged enough things that weren't easy to reproduce to know that little details like that could be important.
yep. that can be important!
-
they may be concurrency safe between different users, but not against a single user with multiple concurrent sessions
It may just be me, but it seems that that would take effort to cause; a design that can handle multiple concurrent users should be able to handle one user in several locations at once just as well.Then again, this is Dicsource…
-
It may just be me, but it seems that that would take effort to cause;
Yes...because the bugs around here are always simple.
-
I SWEAR IT WASN'T ME!
-
More cooties incoming ... while I move us to SSL ... hold tight
@pjh @apapadimoulis upgrading to SSL now
-
Wasn't me either; none of mine were auto-reading
-
TDWTF 100% more SSL 100% more SPDY
-
Coulda given us a bit more warning!
-
Yeah, it took me two F5's to get everything working again.
-
TDWTF 100% more SSL 100% more SPDY
{ [Error: unable to verify the first certificate] code: 'UNABLE_TO_VERIFY_LEAF_SIGNATURE' }
?
-
Is it me or is it much more common for websites to go down nowadays?
Reddit goes down all the time. Imgur has been down half the time these last few days. TDWTF goes down. Steam often goes down for no reason. Twitter allegedly goes down pretty often too (I don't really use it).
Shit, servers, get it together!
-
whats the repro?
-
using nodejs with require package
request('https://what.thedailywtf.com/', function(err){ console.log(err); });
i can fix it on my end but it doesn't like something about that certificate that's gong to require me to turn off cert validation on my end.
-
i can fix it on my end but it doesn't like something about that certificate that's gong to require me to turn off cert validation on my end.
Did you add the certificate chain as well? Comodo has an intermediate certificate that is required in some browsers.
Edit: The intermediate file would have been in the zip file and probably called what_thedailwtf_com.ca-bundle
-
Is it me or is it much more common for websites to go down nowadays?
It is not just you. Google and Facebook seem to be the only ones who can keep shit running for any length of time.
Shit, servers, get it together!
Don't blame the hardware. It is the assholes writing the software for it that is the problem. We have seen the enemy, and it is us.
-
Did you add the certificate chain as well? Comodo has an intermediate certificate that is required in some browsers.
that would probably be it.
i fixed socksite by just disabling the cert check:
Sockbot instances will also need to be fixed. fis for that is incoming shortly
-
Yeah I need to make a pem and add an after rule to add it, I wonder if NGINX is a stickler about file formats.
-
Sockbot instances will also need to be fixed. fis for that is incoming shortly
I've done the URL fix already
-
On the upside, this does cut down on bot traffic :)
Got to take my mum for Dinner, will check back a bit later and figure out how to get a pem into here so you have all the intermediates
-
I've done the URL fix already
and for some rasin the unauthorized fix is already in there.... i wonder why?
-
No idea; I'm getting leaf sig errors on my bots
-
So @zoidberg won't respond to a summons at the moment?
-
I've performed a few mercy killings.
-
No idea; I'm getting leaf sig errors on my bots
hmm that's odd.
So @zoidberg won't respond to a summons at the moment?
apparently he will. ;-)
-
hmm that's odd.
I know what the issue is; committing a fix nowEdit: Yep, was exactly what I thought; had to fix my workaround for that silly issue in
request 2.55