Setting up Bookwyrm
As a book enthusiast and long time GoodReads user, I really liked the idea of BookWyrm, especially to get rid of some of the shortcomings like poor performance and to be able to limit the audience just to be my closer circle.
|I know that this isn’t the idea of social networking and I probably lose good suggestions from other users, but still.
How to run it?
Setting up BookWyrm is quite easy: The default running mode is via docker-compose which works perfectly fine on my local machine. When I try the exact same steps on my VM, it also fetches the different containers (we will come later to that), fires them up until Postgres fails to bind to its default port.
Problems with Postgres
How to get it running?
I haven’t checked in great detail why this happened, maybe the default configuration is to bind
0.0.0.0 or something within the network config of docker-compose is amiss.
For my use case I’d rather be able to use the instance of Postgres that is already running on
my VM, just to save a bit of resources.
There is no easy way to change the configuration to another database host (or container for that matter), so I had to force my way through and did the following:
After that, the the database came up perfectly and docker-compose exited with success.
Which version exactly?
Next up on the list is to run the database migration to get the initial data in place.
bw-dev script presented me this goodie:
Applying bookwyrm.0077_auto_20210623_2155...Traceback (most recent call last):
File "/usr/local/lib/python3.9/site-packages/django/db/backends/utils.py", line 82, in _execute
psycopg2.errors.SyntaxError: syntax error at or near "FUNCTION"
LINE 26: FOR EACH ROW EXECUTE FUNCTION book_trigger()...
This one is probably on me: Apparently, BookWyrm uses features from Postgres v11, which is a problem when you are still running v9. Debian or rather apt-get is quite careful, when it installs a new version of your database and you might end up with more than one:
$ su - postgres -c 'pg_lsclusters' # Ask for existing clusters
Ver Cluster Port Status Owner Data directory Log file
9.6 main 5433 online postgres /var/lib/postgresql/9.6/main /var/log/postgresql/postgresql-9.6-main.log
11 main 5432 down postgres /var/lib/postgresql/11/main /var/log/postgresql/postgresql-11-main.log
Considering the output, it looks like there are two clusters configured and the older one is running, so we just have to migrate the existing one to v11:
$ su - postgres -c 'pg_dropcluster --stop 11 main' # Delete the empty default cluster
$ su - postgres -c 'pg_upgradecluster 9.6 main' # Upgrade the other one
After verification everything is running properly, I just got rid of v9.6 entirely:
$ su - postgres -c 'pg_dropcluster --stop 9.6 main'
$ apt autoremove postgresql-9.6
$ apt autoclean
The next run of
./bw-dev initdb went a bit further until it failed with this error:
psycopg2.errors.InsufficientPrivilege: permission denied to create extension "pg_trgm"
I am not a big fan of using triggers for business logic, because they make upgrades difficult, kind of hide logic and don’t necessarily convey how something is done. Nevertheless, the easiest way here was to temporarily elevate the privileges of the database user:
alter role user_name superuser;
alter role user_name nosuperuser;
So the real question is: Am I going to use it?
I must admit that depends on some things I have to consider:
Although I understand the social aspect, I’d really prefer to be able to limit the instances it can connect to without blocking them directly. (This is something that doesn’t scale well.)
Imports from GoodReads aren’t up to par, when I initially imported 96 books which took quite a while, 23 couldn’t be found. This is remarkable, both in the failed in percent and also, if you consider the export from GoodReads actually contains the ISBN.
The complexity of the stack if kind of frightening, I am not sure the the numbers on the instances will ever justify the Celery task queue with Redis as backend. (Scaling it when necessary would have been the better approach, imho.)
docker-compose shouldn’t be the default mode. Maybe I will really start to take it apart and create a documentation how to do that on the way.
You can find my instance of BookWyrm here: