Closed clmnin closed 3 years ago
Confirmed. The crash is due to the listen-addresses port change in the last step of new instance bootstrap. Restarting the container on the same directory should work.
A better workaround is to use the edgedb server
CLI to configure and start an instance in a docker container:
edgedb server init a7docker --version=1-alpha7 --method=docker
# connect to repl
edgedb -I a7docker
@elprans Thank you. It worked.
I was previously using the verbose version
pool = await create_async_pool(
host=EDGEDB_HOST,
database=EDGEDB_DB,
user=EDGEDB_USER,
)
Never occurred I could use the json from .edgedb/credentials/
to connect to an instance with ease.
pool = await edgedb.create_async_pool('my_instance')
Why was the port changed (from 5656 to 10702)?
Why was the port changed (from 5656 to 10702)?
When you run edgedb-server
directly, the default port is still 5656
. However, instances created by edgedb server init
select a random port, because there may be many of them, and the idea is that you'll be relying on the instance parameters read from .edgedb/credentials
.
This should not be closed. Docker image still fails. It's very inconvenient to wait for container to fail and then start again before we can use it. Can we update the image? Or should we move this issue to edgedb-docker repo?
This is an edgedb-docker issue, and yes it's actionable.
This looks like have been fixed with backend connection pool implementation, so works on nightly. Original error look like a crash because connection to postgres was interrupted. At least running the following works fine:
docker run -it --rm -p 5656:5656 edgedb/edgedb:20210128023945dde634
@elprans, you probably understand older the underlying issue better, so please comment or close the issue.
Yes, @1st1 fixed this with connection pool rework. The original crash was in the internal event sending code.
Steps to Reproduce:
runtime
error