@kinetix Any idea on this?
Always with a plan Stan (@Stan@blimey.social)
Do any other Pleroma admin know how to edit the description of the instance? I’ve noticed that Fediverse Observer shows the description for Blimey Social simply as “pleroma” and for the life of me,...blimey.social
Kinetix
•What happens if you go to "Edit a server" at the bottom and get in to the editing interface?
Always with a plan Stan
•I have no idea where it’s picking up “pleroma” from.
There’s no description and not even “pleroma” for My Crowd 🤷🏻♂️
Kinetix
•Kinetix
•Always with a plan Stan
•Kinetix
•Kinetix
•Fediverse Observer checks all servers in the fediverse and gives you an easy way to find a home using a map or list.
fediverse.observerMathias Hellquist 🤘 likes this.
Mathias Hellquist 🤘 reshared this.
Always with a plan Stan
•Kinetix
•Always with a plan Stan
•Anyway, are you going to tell me how you updated the description or what? 😜
Kinetix
•Yes, to get a description showing up on fediverse.observer, it looks for a description in one of two places - one, the nodeinfo, which apparently Pleroma breaks spec on (or seems to) by using nodeDescription rather than just Description, so that's part of why f.o doesn't pick up on it. The second place it will look is in the site's meta description in it's index html, which we generally don't have in either pleroma-fe or soapbox, I guess, so I just added it in to mine by hand (please feel free to have a look).
Then, f.o checks for this change and will update it once per day - you can check whether it's done it recently or not at https://fediverse.observer/status where it's reporting on the language check.
Now that you've seen it I should think of another short description for the instance.
Fediverse Observer checks all servers in the fediverse and gives you an easy way to find a home using a map or list.
fediverse.observerAlways with a plan Stan
•Which tags do you use and where do you insert it within the file..?
Well done for working it out!
Kinetix
•Have a loon at the source for mycrowd. It's right near the beginning of the page.
Kinetix
•I think it could be useful for us Pleroma admins to build up a database of tidbits for ourselves, our users. I've had a mediawiki for mycrowd.ca setup for awhile, but really have done nothing with it except document the bot accounts I have locally. I don't think I even mention it in the about page yet.
If you'd like to have a look, it's at https://docs.mycrowd.ca - and I think I have more specific detail on the F.O thing so far than what I mentioned here.
Maybe we can continue to pull & organize tidbits that help that either aren't clear, are missing, or aren't relevant in/to the Pleroma docs.
I wonder if @mathias would be interested, too?
My Crowd
mycrowd.caMathias Hellquist 🤘
•Having said that, I have been tweaking a few basic things and could definitely summarize those up in writing, if nothing else as it’d be good for when I’d ever need to set up another Pleroma instance. If any of that could be helpful for anyone else I should share it too.
Kinetix
•I look forward to hearing further how your tweaks go, the experiences definitely deserve to be shared!
Mathias Hellquist 🤘 likes this.
Always with a plan Stan
•The only html file at the root of my instance/Soapbox was “panel.html”, which is what I added the description to 🤷🏻♂️
Kinetix likes this.
Kinetix
•Always with a plan Stan
•Kinetix
•I've found my description was wiped out, because I did a rebuild of the frontend and wiped it out.
Bleh.
Always with a plan Stan
•Kinetix likes this.
Kinetix
•Always with a plan Stan
•Now we wait…
Kinetix likes this.
Kinetix
•Kinetix
•Always with a plan Stan
•Kinetix
•Mathias - yeah, I always found that input box to be kind of clunky with the type/enter/click kind of routine. I like the new layout better, including having the reasonings for each if one desires, but still not helpful for bulk adds. I guess if you are just using the config files instead of config via db you could bulk import.
That's quite the amazing list of instances you have in there, though. I try to keep mine as short as is reasonable, and remove dead sites, but I think I'm going to need an automated tool to help at some point here.
Kinetix
•Kinetix
•Content warning: re: Pleroma admin talk
I'll leave my thoughts on instance admins trying to set rules for other instances aside for the moment...
I personally would probably try and run phx.server manually from the command line first (or up all the logging configuration to 'debug' in your settings) and see what's causing the 500 errors. That sounds like something that's possibly broken.
I don't think I'd recommend taking a potentially broken setup and then updating it to a different branch. I suppose, though, if you did a full db and pleroma backup, it could be worth trying.
Are you comfortable with git, etc? (honest question) Are you able to spin up a test instance to run through a from-source develop-branch setup?
On my setup, which I wouldn't suggest as being a recommended setup, is that I have directory which is the pleroma code from git, and another directory with pleroma-fe from git. The pleroma-fe bits are symlinked from the pleroma instance/static directory.
I use the adminfe that comes along with pleroma, in it's priv/static directory.
Mathias Hellquist 🤘 likes this.
Kinetix
•Content warning: re: Pleroma admin talk
I honestly don't know what to suggest there as I, uh, am getting what I pay for? heh.. Honestly, don't want to complain when there's so much good stuff that has gone on.
Kinetix
•Content warning: re: Pleroma admin talk
Kinetix
•Content warning: re: Pleroma admin talk
Mathias Hellquist 🤘 likes this.
Always with a plan Stan
•I used to drive the Routemasters on the 38 from the bus garage there 😬
Mathias Hellquist (Friendica)
1. I got a message saying: "warning: the dependency :earmark_parser requires Elixir "~> 1.10" but you are running on v1.9.1". Uhu? So why wasn't the correct version installed, as I followed the official installation instructions? Is this something I need to fix?
2. Got another message a couple of times: "Rebar3 detected a lock file from a newer version. It will be loaded in compatibility mode, but important information may be missing or lost. It is recommended to upgrade R... show more
1. I got a message saying: "warning: the dependency :earmark_parser requires Elixir "~> 1.10" but you are running on v1.9.1". Uhu? So why wasn't the correct version installed, as I followed the official installation instructions? Is this something I need to fix?
2. Got another message a couple of times: "Rebar3 detected a lock file from a newer version. It will be loaded in compatibility mode, but important information may be missing or lost. It is recommended to upgrade Rebar3". So...is the lockfile part of the git repo? Do I have a newer one, or not? If I do, do I really need to update it? Or is it updated? How do I update it?
These are some of the questions I have as my db_backup still trots along and I'm waiting eagerly to see if it all has worked (and which obviously is the reason I'm typing this from my Friendica account right now). :)
Mathias Hellquist (Friendica)
Right. So having re-installed it all afresh I also tried adding a source to get past the error messages. Now I *think* I have version 1.25 of what it tells me I should have version 1.10. However, this seems to lead to compilation problems for Hackney instead. I get a million of these errors now (and it doesn't federate):
´´´
20:05:03.166 [error] #PID<0.1968.0> running Pleroma.Web.Endpoint (connection #PID<0.1967.0>, stream id... show more
Right. So having re-installed it all afresh I also tried adding a source to get past the error messages. Now I *think* I have version 1.25 of what it tells me I should have version 1.10. However, this seems to lead to compilation problems for Hackney instead. I get a million of these errors now (and it doesn't federate):
´´´
20:05:03.166 [error] #PID<0.1968.0> running Pleroma.Web.Endpoint (connection #PID<0.1967.0>, stream id 1) terminated
Server: moc.d-x-b.com:80 (http)
Request: POST /inbox
** (exit) an exception was raised:
** (UndefinedFunctionError) function :hackney_bstr.join/2 is undefined (module :hackney_bstr is not available)
(hackney 1.18.0) :hackney_bstr.join(["", "users", "cmegane"], "/")
(hackney 1.18.0) /opt/pleroma/deps/hackney/src/hackney_url.erl:140: :hackney_url.normalize/2
(hackney 1.18.0) /opt/pleroma/deps/hackney/src/hackney.erl:311: :hackney.request/5
(tesla 1.4.1) lib/tesla/adapter/hackney.ex:71: Tesla.Adapter.Hackney.request/5
(tesla 1.4.1) lib/tesla/adapter/hackney.ex:33: Tesla.Adapter.Hackney.call/2
(pleroma 2.4.4) lib/pleroma/object/fetcher.ex:227: Pleroma.Object.Fetcher.get_object/1
(pleroma 2.4.4) lib/pleroma/object/fetcher.ex:200: Pleroma.Object.Fetcher.fetch_and_contain_remote_object_from_id/1
(pleroma 2.4.4) lib/pleroma/web/activity_pub/activity_pub.ex:1569: Pleroma.Web.ActivityPub.ActivityPub.fetch_and_prepare_user_from_ap_id/1
´´´
Mathias Hellquist (Friendica)
Also a bunch of these:
```
20:00:47.157 [error] beam/beam_load.c(86): Error loading module hackney_bstr:
lambda already defined for label 188. To fix this, please recompile this module with an OTP 25 compiler.
20:00:47.234 [error] Loading of /opt/pleroma/_build/prod/lib/hackney/ebin/hackney_headers.beam failed: :badfile
```
Kinetix
•Thing is, I have no idea where pleroma's mix and build setup is pulling rebar3 from, but it seems to pull in a broken rebar3 (or at least, it was).
I was able to fix that by grabbing rebar3 from github, building it myself and putting it in my user's ~/.mix folder.
I wonder if that part had anything to do with whatever was going on at pleroma's gitlab where it was producing things like aarch64 binaries when requesting amd64? I can't imagine the potential mess that could have gone on because of that. As I don't know where the rebar3 was coming from, I can only guess.
It sounds like you've got more of a mess going on there, though, I wish I could provide more insight, but perhaps my recent trials will help?
Mathias Hellquist (Friendica) likes this.
Mathias Hellquist (Friendica)
Yeah, sorry, I am posting those updates mostly for my own sake if nothing else, so I get a record of what happens, but with the added benefit that one of you guys possibly go "hey! THAT thing can be fixed easily with the mix.clamsteamer" or similar.
Turned out doing a "sudo -Hu pleroma mix local.rebar" got me past the rebar woes. Took me a couple of hours to learn that though, haha. :)
I have actually wiped it all again now and am trying to get it up again, but with existing (previously imported) database. Not doing well right now, but hey, I'm getting closer for each attempt I think. :)
Kinetix likes this.
Mathias Hellquist (Friendica)
Ubuntu 20.04, nginx (latest whatever). Just a bog standard install of a VPS. Apart from adding the Erlang stuff to sources it is the default Ubuntu sources. I say that as I am thinking the instructions possibly should include pointers to sources that aren't actually part of the defaults to for example Debian/Ubuntu (which I would consider is the most common Linux distro still).
Even after my clean installs I get lots of these too, saying "Error: n.text is undefined" which I get pretty much always when trying to post either an original post or a reply to a post. Usually the pos... show more
Ubuntu 20.04, nginx (latest whatever). Just a bog standard install of a VPS. Apart from adding the Erlang stuff to sources it is the default Ubuntu sources. I say that as I am thinking the instructions possibly should include pointers to sources that aren't actually part of the defaults to for example Debian/Ubuntu (which I would consider is the most common Linux distro still).
Even after my clean installs I get lots of these too, saying "Error: n.text is undefined" which I get pretty much always when trying to post either an original post or a reply to a post. Usually the posts are actually posted anyways, but just to make sure I always have to type the post in another tool to ensure I won't lose it.
Also I get lots of 500 error still. Reminder: this is after completely wiping the folders, re-downloading them from stable branch and re-compiling them. You can see the messages in the screenshot. The messages are not related, they just happened to be on the screen at the same time for the screenshot but usually happen regardless of each other (just a lucky screenshot).
Kinetix
•Oh, on the rebar point specifically, 'mix local.rebar' was what was making me believe Pleroma pulls rebar in from somewhere, as it would continue to replace rebar with another complaining one. (I am running Erlang/OTP 25, and my Elixir was compiled with it, so it was kind of entertaining being told it needed OTP 25 - which I think I missed mentioning in my last response)
Kinetix
•Mathias Hellquist 🤘
•Haha, yes, well…:) I was dragging my feet on updating here. So…after some fiddling about I managed to do the re-install (which was a complete wipe and a restore of db). Turned out it didn’t actually fix anything, which had me worried. Still lots of error messages of different kinds and CPU usage, which had gone down from a solid 100% was still frequently (like, most of the time) around 70-80%, with only occasional dips down to 30-ish%.
When I compared it to my Friendica instance, which in all honesty “shouldn’t be as efficient” as Pleroma, the Friendica server was/is rarely over 10% CPU. I concluded something still wasn’t right.
Then I decided to do another wipe, but this time instead go for the OTP release. Theory being “if I compile badly, or with errors, get something that is already compiled instead”. It could be pointed out that I have an “amd64”. Anyways, I switched to t... show more
Haha, yes, well…:) I was dragging my feet on updating here. So…after some fiddling about I managed to do the re-install (which was a complete wipe and a restore of db). Turned out it didn’t actually fix anything, which had me worried. Still lots of error messages of different kinds and CPU usage, which had gone down from a solid 100% was still frequently (like, most of the time) around 70-80%, with only occasional dips down to 30-ish%.
When I compared it to my Friendica instance, which in all honesty “shouldn’t be as efficient” as Pleroma, the Friendica server was/is rarely over 10% CPU. I concluded something still wasn’t right.
Then I decided to do another wipe, but this time instead go for the OTP release. Theory being “if I compile badly, or with errors, get something that is already compiled instead”. It could be pointed out that I have an “amd64”. Anyways, I switched to the OTP, which was an improvement. Still higher CPU than the Friendica instance (which actually runs on an ARM CPU), but a less than previously, going between 15-50% CPU on average, with only odd spikes that went higher.
Whilst searching for things to solve this I came across Akkoma, and this was also the reason I was dragging my feet in telling you guys about this (as there seems to be some conflict between various dev groups, leading to a split+fork etc). You see, being on an OTP release I figured I had done the (not-so) “tricky” part already with moving config files, upload folders etc, so testing Akkoma OTP shouldn’t be overly difficult. If nothing else, if it didn’t change anything it should be easy to go back.
So that is basically where I am: my server is now running Akkoma. Average CPU usage: 3-5% (like I remembered the Pleroma instance had when I installed it first time, on a v1.x something). It all “just works”. Akkoma has a bunch of other additions that I don’t care at all about, it was easy enough to tailor to look/feel like Pleroma-as-I-remember-it but with a few nifty additions.
Now I’m sitting here wondering if I should update Postgresql from v12 to perhaps v14, and if it is do-able without major surgery, but also, if I “have to” or if “I should due to unknown benefits”.