@Undo @ArtOfCode All SD instances appear down for the last hour, or at least not pinging MS. Mine is complaining about not being able to login to meta.stackexchange.com with max retries exceeded. Nothing has changed in my config other than normal pulls via !!/pull or MS for quite some time.
[01:02:27.991] Login error LoginError: failed to get `acct` cookie from Stack Exchange OpenID, check credentials provided for accuracy
[01:02:28.302] Login error LoginError: failed to get `acct` cookie from Stack Exchange OpenID, check credentials provided for accuracy
[01:02:28.614] Login error LoginError: failed to get `acct` cookie from Stack Exchange OpenID, check credentials provided for accuracy
[01:02:28.615] Exception: Failed to log into stackexchange.com, max retries exceeded
2019-04-14 01:02:28.614952 UTC
The credentials in the config file still work to log into SE. I'm able to access the SE API from my EC2 instance.
[01:43:11.934] Logged in using cached cookies
[01:43:14.845] KeyError: 'time'
2019-04-14 01:43:14.844176 UTC
File "ws.py", line 139, in <module>
chatcommunicate.init(username, password)
File "/home/ec2-user/SmokeDetector/chatcommunicate.py", line 112, in init
join_command_rooms()
File "/home/ec2-user/SmokeDetector/chatcommunicate.py", line 132, in join_command_rooms
room.join()
File "/usr/local/lib/python3.6/site-packages/chatexchange/rooms.py", line 51, in join
return self._client._join_room(self.id)
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, pattern-matching product name in body, pattern-matching product name in title, pattern-matching website in body, +3 more (492): What is The Keto Diet Pills? by Pure Life Keto on stackoverflow.com
@SmokeDetector This appears to have not had the flag raised that's supposed to be automatically sent when the post is deleted with flags from MS on it and there's non-TP feedback.
@SmokeDetector I suspect this was deleted by the user. While the question was deleted, both my flag and SD's flag were still pending. I've retracted both flags.
[ SmokeDetector | MS ] Url in title, bad keyword in body, bad keyword in title, body starts with title and ends in url, potentially bad keyword in body, +1 more (390): suppleement.com/keto-ignite/ by rezima1 on datascience.SE
@ArtOfCode @Undo The problem that we're having with SD instances crashing is that the Socket Science room on chat.SO has been deleted. This is causing all SD instances which are at a version that includes that room in rooms.yml to crash.
I've raised a flag asking for the room to be undeleted, as that's the easiest way to get the instances that are not currently responsive to become functional.
OK. Other instances, once they revert far enough back, should be able to be !!/pulled to the current version.
Once the Socket Science room on chat.SO is undeleted, I'll re-revert this most recent change (which was a revert of the change moving Socket Science to chat.SO).
@iBug The problem was that the Socket Science room on chat.SO was deleted. You can pull a current version of the GitHub repository, which should work. Currently, rooms.yml was changed back to having Socket Science on chat.SE. I've asked for the chat.SO room to be undeleted and will change rooms.yml back once that happens.
@ArtOfCode Socket Science on chat.SO has been undeleted. There's now a message from another user in the room, so we should have a bit more margin prior to it being deleted again. We should keep the room active enough not to be deleted. The method I've used in other rooms is to ask to have it added to the rooms which Generic Bot posts keep alive messages to. Would you like to use that, have SD post into it often enough, or something else?
@ThomasWard There was an issue that resulted in all SD instances not being able to function. It's been resolved, but Solar doesn't appear to have recovered. It may recover on its own by the time you see this, or it may need to be kicked.
@Makyen That pattern looks like it's already caught by Pattern-matching email in body and Pattern-matching email in answer; append -force if you really want to do that.
@Olivia That pattern looks like it's already caught by Bad keyword with email in answer and Bad keyword with email in body; append -force if you really want to do that.
we could apprach this differently, simple addition is really easy to implement and understand but you would really like for there to be some sort of boosting effect when you have a lot of reasons
@double-beep the recommended weight is somewhere around 200-300 IIRC
it is not enough to avoid a large number of false positives for completely innocent stuff like repeated characters + apparent phone number in title (could be an IP address or an error code or what not)
@tripleee having blacklisted user reason and potentially bad keyword looks very suspicious. Consider the case where a user spammed a website, the website was watched and now, except for the reason 'blacklisted user', there is 'potentially bad keyword'.
but actually increasing the weight of blacklisted user reason is not something that should be done...
Similarly, a watched site is intentionally not meant to mean much - it is simply to remind us that this has been seen before, not to tell us that it's probably spam (unlike the blacklist). Same story with bad ASN.
@double-beep no, but occasionally someone will incorrectly give tpu- as feedback
@angussidney Not testing. That's a moderately common failure reason for manual flags. It's often because of a race condition where the deletion of the post is not yet known by FIRE (e.g. the deletion happened after the popup is open and the notification hasn't come through on the MS WebSocket yet).
[ SmokeDetector | MS ] Blacklisted website in body, mostly non-latin body, mostly non-latin title, pattern-matching website in body, potentially bad keyword in body (363): انا وصاحبي وامو و خالتو 2 by user165049 on wordpress.SE
