I'm not sure if it joins the room prior to its first message. When it joins the room, may vary bassed on what it's doing in the room, with, presumably, just reporting into the room being the lowest priority for when it needs to join. I know that how dependent SD is on the room's existence depends on what it's doing in the room.
@Makyen That pattern looks like it's already caught by Pattern-matching website in body, Bad ns for domain in answer, Pattern-matching website in answer, and Bad ns for domain in body; append -force if you really want to do that.
@Glorfindel That doesn't look like a valid user URL.
@Makyen That pattern looks like it's already caught by Potentially bad keyword in answer and Potentially bad keyword in body; append -force if you really want to do that.
@iBug DeepSea continued to have problems. A couple of times it encountered the out of memory issue when doing a !!/watch. That was able to be worked around by retrying the command, but it was confusing to people. It was also experiencing issues with disconnecting from watching rooms for commands. We had to !!/reboot multiple separate times when it got disconnected from either CHQ and/or the Tavern.
[ SmokeDetector | MS ] Potentially bad asn for hostname in body, potentially bad keyword in body, potentially problematic ns configuration in body (3): Aplikasi Cheat Bandar Ceme by CheatPasti on english.SE
@EliahKagan If you're sure something is spam, you can go ahead and use !!/report, which will also scan it, but will report it even if the scan doesn't see anything.
@terdon No, it can't see deleted posts. It'd be nice (undel, report, del), but probably not worth it at the moment. I got the domain out of it and watched it, which is probably the most important thing.
@terdon OTOH that particular (newer) post seems to have been deleted in an unusual way. Normally posts that are deleted as spam are locked, but that one isn't. Maybe it would be worthwhile to undelete and delete it, because of that. I don't know.
@iBug Halflife and PulseMonitor also occasionally suffered from something similar, I'm not altogether sure what goes wrong but Docker on EC2 (I guess you are using, too?) seems to have some issues
One of my school servers got upgraded to Buster (was Stretch) and sudo-ldap just never respects the rule sudoOption: !authenticate from our LDAP server. All other machines are behaving correctly (we run predominantly on Debian with a small number of CentOS).
That very machine ran into another issue while trying to install Debian Buster - namely the installer detected the NIC, but couldn't work with it. It was particularly confusing that it just detected the NIC and gave strange errors whenever it tries do anything on it.
@AJ that's been my hypothesis, I don't think I have seen any samples which were not #drugs
it's of course not impossible that the same tactic could be used by others going forward but let's concentrate on #drugs in any event for now, they should be the vast majority
@tripleee I ran a simple query to find posts tagged as #drugs. SELECT * FROM p_posts where find_in_set(1, p_posts.tags); It returning zero rows. Is my query wrong or there is no post tagged as such?
> Pattern-matching website in body, pattern-matching website in title ---------- Pattern-matching website in body - Position 0-8: football Pattern-matching website in title - Position 0-8: football
https://github.com/Charcoal-SE/SmokeDetector/blob/master/findspam.py#L1903 lists "football" as a website pattern but should it not be seen in a URL to qualify ...?
Searching username ^[a-zA-Z0-9]{4,10}2ch$ (or perhaps, not so safe, 2ch$) on MS gives interesting results. Should that expression be blacklisted? It'll increase the weight of some posts, so they'd be autoflagged with 4 flags.
@iBug I'd be quite careful with this one. It looks generic enough that there would be considerable FP. For the ones that match the primary pattern (?-i:^[A-Za-z][a-z]+row$), there are none below the default autoflagging threshold in the last 5 months. For the ones that were below the default, we have to go back almost 9 months to find one that wasn't actually autoflagged.
@double-beep At an absolute minimum, things like that need to spend some time on the watchlist prior to blacklisting. While that one isn't too bad, and is not too likely to have FP, we need to verify that. The data on MS is very biased towards things being spam.
@iBug Feel free to do so. Mine is having an issue that it's not reliably pulling and restarting after updating the watchlist (probably blacklist too). A manual !!/pull is often needed. It's correctly adding things to GitHub, but it's not taking the update itself. I need to track down why that's the case, but I'm not going to be able to work on it for a few/several hours. So, if yours is good to go, particularly if you are going to be around to solve any problems, let's use it.
[ SmokeDetector | MS ] Url in title, bad keyword in body, bad keyword in title, body starts with title and ends in url, pattern-matching website in body, +6 more (689): healthcircle365.com/keto-crush/ by caymedrick on wordpress.SE
Everyone: Lunar Eclipse is up on new Infra. I have to do some testing to see whether there's some evil going on here, but it's entirely possible I will have to spin a second IP on the box running the Smokeys I have now, and route it over another IP Outbound. That'll take some time, so Solar will be offline until I get that figured.
The answer will be revealed in my documentary on the origins of A New Hope and the unsung heroes behind it all based on My book Cinema Alchemist on making A New Hope and a blow by blow account behind the scenes.
Roger Christian
@ThomasWard I'm having some difficulties telling the differences between LXC and LXD since they're made by the same organization and is similarly names.
I'm running a few LXC containers on an LVM volume as "light-weight VMs".
@tripleee That pattern looks like it's already caught by Pattern-matching website in answer and Pattern-matching website in body; append -force if you really want to do that.