haccfiles/docs/services/hedgedoc.md
stuebinm 41d82ae436 meta: new structure
we decided to:
 - get rid of unused packages
 - simpify the directory layout since we only have one host anyways
 - move our docs (such as they are) in-tree
2024-01-11 23:49:26 +01:00

1.8 KiB

+++ title = "hedgedoc" taxonomies.categories = [ "services" ] +++

hegedoc was once called codiMD, so container, config and users are still called codimd.

Do NOT change this unless you're sure what you're doing.

We have two instances:

  • pad-hacc/pad.hacc.space is connected to our SSO/uffd
  • pad-i4f/pad.infra4future.de is not connected to our SSO and meant to be more public

Basic Troubleshooting

Usually if hedgedoc dies, it's because postgresql wasn't there yet. Just restart hedgedoc.

More Troubles

log into the container and take a look at the logs

sudo nixos-container root-login codimd
journalctl -e

fixing failed database upgrades

see https://docs.hedgedoc.org/guides/migration-troubleshooting/ (copied below for convenience?):

In some cases, HedgeDoc might apply migrations without correctly saving the progress. It will then refuse to start with "already exists"-errors like ERROR: type "enum_Notes_permission" already exists.

Get the name of the failing migration and append .js to it. For example, if you encounter this error:

== 20180306150303-fix-enum: migrating =======

ERROR: type "enum_Notes_permission" already exists

the name of the failed migration would be 20180306150303-fix-enum.js.

The SQL-statement may look like this:

INSERT INTO "SequelizeMeta" (name) VALUES ('20180306150303-fix-enum.js');

Make sure HedgeDoc does not run and insert the name into the SequelizeMeta table. Enter the container switch to the postgres user, open psql and commect to the codimd database:

su postgres
psql
\l
\c codimd
UN adjusted SQL STAMEMENT from above ]
\q

Start HedgeDoc again and observe if it starts correctly. It may be necessary to repeat this process and insert multiple migrations into the SequelizeMeta table.