Notices where this attachment appears
-
Embed this notice
Also I regret naming that machine "fedilistdb" because that hasn't been what it does for at least a year and a half. "zanzibar" is a Metal Gear joke. Those are the public machines. (And, to differentiate between downtime at hosting companies and the local net fucking up, I tossed in Level 3 and Google.)
The TPi2 that FSE is attached to is named "arsenal" and FSE itself lives on the host named "ray"; I chickened out of naming all of the boards attached to it "ray-a01e", "ray-a05e", "ray-b01f", "ray-c01h" (which would fit the MGS2 joke better but would also be harder to remember), so the other three systems on it are named "rex" (fedilist lives half there and half on flcb now, though the whole thing will be on flcb later; have some code to write and then a bunch of wireguard shit to rejigger), "peacewalker" (because it is the large one), and "tx55" (just a CM-4). (Some of these names might look familiar if you saw the Revolver Grafana dash: http://pbuhwjjhrzcvrghtfqlwqlgabuzc7jnkqv4swekxjvv7pgubd7jjoiqd.onion/public-dashboards/8643a3beee524385bd0ef6a304dbbd65?orgId=0&refresh=6s )
When FSE downtime is *not* caused by local ISP horseshit (or after I move FediList), you can check its status page: https://fedilist.com/instance/recent-changes?host=fsebugoutzone.org . There is an RSS feed linked there as well, so you can get alerted when its up by shoving that into your RSS reader or bashing out a script that talks to RSS feeds.
An undocumented feature: the "host" field there is comma-separated. So you can actually watch multiple systems at once: https://fedilist.com/instance/recent-changes?host=fsebugoutzone.org%2Cdarkdork.dev%2Cannihilation.social . So you can use that to get an RSS feed (in this case https://fedilist.com/instance/recent-changes/rss?host=fsebugoutzone.org%2Cdarkdork.dev%2Cannihilation.social ) and get status monitoring for as many instances as you feel like.
-
Embed this notice
@ins0mniak @ThatWouldBeTelling @dsm Yeah, it was doing that on my old router. I ended up with a new router eventually but I ended up shipping it off to a Frantech node.
I'm running a bunch of Tor services, like this feller: http://pbuhwjjhrzcvrghtfqlwqlgabuzc7jnkqv4swekxjvv7pgubd7jjoiqd.onion/public-dashboards/8643a3beee524385bd0ef6a304dbbd65?orgId=0&refresh=6s
-
Embed this notice
This all seems to be working fine, I am pleased.
> uploading is faster than it *was*
Except that it is more complicated than that: the person doing the upload now has to wait for the ingestion to occur, whereas before it would happen on-demand, so there was no waiting. On the other hand, the amount of time from "upload initiated" to "first byte reaches first client" is lower (in terms of both wall-clock time and the actual amount of time required by the machine).
Gotta do some writing, gotta do some hacking on the network code, gotta stand up a new box (thanks, @graf) for FediList, also the Grafana dash for Revolver is back up: http://pbuhwjjhrzcvrghtfqlwqlgabuzc7jnkqv4swekxjvv7pgubd7jjoiqd.onion/public-dashboards/8643a3beee524385bd0ef6a304dbbd65?orgId=0&refresh=6s . If you have drugs, please bring them by; I have too much shit to do to ever sleep again.
-
Embed this notice
@NonPlayableClown @DiamondMind @lainbot @lanodan @toiletpaper Yes. Like, multiple times daily. Constantly. Usually there is another binary deployed by the time the one currently running crashes.
I thought that never shutting up about Revolver was a tip-off, but all day, every day, except that I sometimes eat or sleep, I am either hacking on Revolver or trying to make some money.
Currently I am trying to import bae.st's media, this was slow until I realized the obvious: I can just shit blocks all over the place because it's distributed storage anyway, so I'm basically flooding every disk that has enough space with 3TB of waifus and this is stressing basically every instance but it's 23.2% done now, plus the main media server is doing sort of an on-demand data migration.
The travails are still visible on the Apotheosis stream ( https://fsebugoutzone.org/apotheosis ) and on the Grafana dashboard ( http://pbuhwjjhrzcvrghtfqlwqlgabuzc7jnkqv4swekxjvv7pgubd7jjoiqd.onion/public-dashboards/8643a3beee524385bd0ef6a304dbbd65?orgId=0&refresh=6s ), both of which are linked from the top of the page here: https://blog.freespeechextremist.com/blog/apotheosis.html . Shit's gonna be broken a lot, but especially right now, there's a lot of congestion in the system.
-
Embed this notice
@lamp @EdBoatConnoisseur @neo
> It doesn't matter what people do with it when it's clearly and explicitly unfinished and unreleased, it's entirely their problem.
Except that it is network software and it will be the network's problem.
> Why should I donate when I can't see what's being done?
You don't have to.
> It may as well be a scam.
It has been serving the media for FSE for a long time, it's stepped in for objects when Pleroma had that accidental MRF DDoS bug, it's currently doing some horrible forked /inbox thing so I can compare its responses to Pleroma (so you may see it in your logs when it fetches actors that it hasn't refreshed in a while, for example), the ingestion process is being live-streamed (sorta, via https://fsebugoutzone.org/apotheosis and the Grafana dash at http://pbuhwjjhrzcvrghtfqlwqlgabuzc7jnkqv4swekxjvv7pgubd7jjoiqd.onion/public-dashboards/8643a3beee524385bd0ef6a304dbbd65?orgId=0&refresh=6s ). It would be more work to make a scam (and if it were a scam, I'd be pushing donations harder).
-
Embed this notice
@teto @hj @i @j @j @lanodan @shitpisscum Actually, if you have Tor then it's viable on the grafana dashboard: http://pbuhwjjhrzcvrghtfqlwqlgabuzc7jnkqv4swekxjvv7pgubd7jjoiqd.onion/public-dashboards/8643a3beee524385bd0ef6a304dbbd65?orgId=0&refresh=6s .
baest_media_ingestion.png