In a conversation with another user (in Portuguese), he mentioned that:

"(He is) commenting from an alt account (Friendica) because (my) instance, the/brain/bin, is returning an error when (he) tries to pull the thread through Sharkey."

I asked if he got any specific errors and now I'm waiting for a reply. But in the meantime, I tested from my side, and the profile he mentioned (@dsilverz@calckey.world) gives a 500 Internal Server Error when trying to open it, but searching shows the profile (though opening the profiles still results in a 500 error).

The domain isn't defederated or dead as per the local federation page either.

Is it a known issue? If not, anything else I could help with?

Thanks in advance.

EDIT:

Checking the user's alt's comments, he mentioned 7 months ago about issues between federation between calckey.world and Fedia.io. Since Fedia.io is also powered by Mbin, maybe it's an issue between site engines?

@Auster@thebrainbin.org @askTheBrainBin@thebrainbin.org I received this post through notification (so, the MBin notification did arrive at my Calckey.world side). Not sure if this reply of mine will make it through, though. This is what happens on my side whenever I try to fetch Auster's comments: (the screenshot is embedded, but its transcription follows)Response is invalid: it could communicate with the server, but the data obtained was incorrect.I also can't see Brain Bin profiles from my side. I only see Auster's profile avatar likely because it was part of the notification payload sent from Brain Bin to Calckey.As Auster noticed, I had the exact error when trying to fetch other MBin instances. Perhaps it's not something to do with The Brain Bin specifically, but with MBin in general. Or, as I noticed when I first accessed Brain Bin as guest, the Anubis anti-bot mechanism may be affecting certain ActivityPub endpoints (for example, Calckey expects an ActivityPub JSON but would be receiving an Anubis challenge instead).Doesn't seem like an issue with Calckey.world because I've been successfully interacting with lots of different Fediverse instances and platforms: Lemmy, PieFed, Mastodon, PixelFed, GoToSocial, etc. Also, Calckey.world got recently updated (I saw an update notification today's morning after Misskey updated a critical vulnerability issue that was discovered this week, also releasing the update to Misskey forks which includes Sharkey and Calckey, so Calckey.world is up-to-date).

@Auster@thebrainbin.org @askTheBrainBin@thebrainbin.org Update: my reply to this thread managed to make it through (so both instances are communicating somehow, despite the fetching issues), but I see that my comment appears as "NSFW" over there (because the thread title is being parsed as Content Warning, and Calckey replies inherit the parent post's CW by default). I'm trying to send this additional reply where I removed the default CW, in case initial reply was filtered due to CW and went un-notified.

I don't know what the exact issue is, but if it was the anubis setup, then not only calckey would have such problems. I believe it may be the accept header that calckey is sending, but I don't know which one that is

I looked through their code and it seems like the get request accept header looks like this: application/activity+json, application/ld+json; profile="https://www.w3.org/ns/activitystreams". Trying to request this thread with curl works without a problem:

curl --header "Accept: application/activity+json,application/ld+json; profile=\"https://www.w3.org/ns/activitystreams\"" https://thebrainbin.org/m/askTheBrainBin/t/1478488/Issue-federating-with-calckey-world

When sending a thread to a server, the user has to be resolved by the other server, only a reference is passed, so it seems like it is working. Why the title is interpreted as a CW when the sensitive property is set to false sounds like a bug to me though

@Auster@thebrainbin.org the error has nothing to do with calckey specifically, only with a mention that is non-unique so in turn throws an error

Should now be solved

@BentiGorlich@thebrainbin.org @Auster@thebrainbin.org Using the curl command line, but with an additional verbose flag (-v) to check the headers, I compared ActivityPub responses from both Brain Bin and Lemmy (a platform with which I interact frequently, rarely experiencing fetching errors), in order to see whether there were any missing key-pairs. As far as I could get, nothing really stood out to me as incorrect or missing. Brain Bin is missing Content-Length the Transfer-Encoding: chunked (expected when content-length is uncertain at request time), but lemmy.world, with which I manage to interact from Calckey without isssues, is also missing both headers, because both Brain Bin and Lemmy.world are behind reverse proxification (Anubis and CloudFlare, respectively) so content-length is unknown to the reverse proxy until the gateway generates the response.So my hypothesis are three.Calckey may be stumbling across some request within the fetching flow where Anubis enters in action (but it's not the profile picture because I successfully fetched an Brain Bin-hosted profile picture from the command line without triggering Anubis, so I have no clue what exactly it would be).Or some ActivityPub key/value is being expected (but again, as far as I looked at it, nothing stood out as missing or incorrect; both payloads, that of Lemmy and that of Brain Bin, are similar).Or maybe the it has something to do with the Calckey IP address (given how we likely both ran the curl requests using domestic IP addresses, while Calckey is likely sitting in a VPS with a VPS IP address; but this would also affect communications going in in the other way, when I successfully got the notification and I managed to send the reply). Or something completely different, I dunno. 🤔