math worker crashes when run in dev mode

[andxblack]

Expected behavior:
Spin up the dev environment using docker-compose and connect to the math worker nREPL as described in math/readme.md

Actual behavior:
polis-math starts up, but exits after about a minute.

polis-math                  | 22-01-14 00:11:35 aefc7c73cc0e INFO [polismath.components.postgres:92] - modpoll 1641255039828
polis-math                  | 22-01-14 00:11:36 aefc7c73cc0e INFO [polismath.poller:23] - Polling :moderation > 1641255039828
polis-math                  | 22-01-14 00:11:36 aefc7c73cc0e INFO [polismath.components.postgres:92] - modpoll 1641255039828
polis-math                  | 22-01-14 00:11:36 aefc7c73cc0e INFO [polismath.poller:23] - Polling :votes > 1641255039840
polis-math                  | 22-01-14 00:11:36 aefc7c73cc0e INFO [polismath.components.postgres:75] - poll 1641255039840
polis-math                  | 22-01-14 00:11:37 aefc7c73cc0e INFO [polismath.poller:23] - Polling :votes > 1641255039840
polis-math                  | 22-01-14 00:11:37 aefc7c73cc0e INFO [polismath.poller:23] - Polling :moderation > 1641255039828
polis-math                  | 22-01-14 00:11:37 aefc7c73cc0e INFO [polismath.components.postgres:92] - modpoll 1641255039828
polis-math                  | 22-01-14 00:11:37 aefc7c73cc0e INFO [polismath.components.postgres:75] - poll 1641255039840
polis-math exited with code 0

To Reproduce:
docker-compose -f docker-compose.yml -f docker-compose.dev.yml up

Device information:
OSX 12.1, Docker desktop 4.3.2 (72729)

[patcon]

Thanks again @andxblack

I've never done development on math (nor have I used the dev compose file), but I'm seeing that @metasoarous added it in July and should have more context :)

[patcon]

Confirmed I'm seeing the same thing.

Logs with docker-compose.dev.yml file and regular compose file both start with:

polis-math              | SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".
polis-math              | SLF4J: Defaulting to no-operation (NOP) logger implementation
polis-math              | SLF4J: See http://www.slf4j.org/codes.html#StaticLoggerBinder for further details.
polis-math              | WARNING: min already refers to: #'clojure.core/min in namespace: polismath.math.clusters, being replaced by: #'clojure.core.matrix.operators/min
polis-math              | WARNING: max already refers to: #'clojure.core/max in namespace: polismath.math.clusters, being replaced by: #'clojure.core.matrix.operators/max
polis-math              | WARNING: min already refers to: #'clojure.core/min in namespace: polismath.math.conversation, being replaced by: #'clojure.core.matrix.operators/min
polis-math              | WARNING: max already refers to: #'clojure.core/max in namespace: polismath.math.conversation, being replaced by: #'clojure.core.matrix.operators/max
polis-math              | WARNING: min already refers to: #'clojure.core/min in namespace: polismath.math.corr, being replaced by: #'clojure.core.matrix.operators/min
polis-math              | WARNING: max already refers to: #'clojure.core/max in namespace: polismath.math.corr, being replaced by: #'clojure.core.matrix.operators/max
polis-math              | 22-01-14 07:38:53 1dca6f073070 INFO [polismath.components.config:152] - >> Starting config component
polis-math              | 22-01-14 07:38:53 1dca6f073070 INFO [polismath.components.core-matrix-boot:39] - >> Starting CoreMatrixBooter with implementation: :vectorz
polis-math              | 22-01-14 07:38:54 1dca6f073070 INFO [polismath.components.postgres:58] - >> Starting Postgres component
polis-math              | 22-01-14 07:38:54 1dca6f073070 INFO [polismath.conv-man:425] - >> Starting ConversationManager

Oddly, when trying to start following the logs with docker-compose logs --follow math, it takes way longer to attach with the docker-compose.dev.yml logs compared to the non-dev one. Not sure what's wrong, so i'll leave it to chris

[metasoarous]

Hi @andxblack. Thanks for reporting this.

This should be fixed as of #962. If it's not, please reopen, or start a new issue if it seems like a different problem.

Thanks again

[nickstares]

Hiya, I am running into this same issue trying to run the project for the first time. I tried to connect to polis-math via nREPL but couldn't connect (even though the nREPL proces did at one point start on 18975) and realized that I also got these messages in the log:

polis-server-1     | info heapUsed: 31745352 heapTotal: 34041856 rss: 95109120
polis-math         | 22-09-05 19:40:32 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :votes > 1661542778121
polis-math         | 22-09-05 19:40:32 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :moderation > 1661542778118
polis-math         | 22-09-05 19:40:32 ee4b3bf5bb6a INFO [polismath.components.postgres:75] - poll 1661542778121
polis-math         | 22-09-05 19:40:32 ee4b3bf5bb6a INFO [polismath.components.postgres:92] - modpoll 1661542778118
polis-math         | 22-09-05 19:40:33 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :moderation > 1661542778118
polis-math         | 22-09-05 19:40:33 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :votes > 1661542778121
polis-math         | 22-09-05 19:40:33 ee4b3bf5bb6a INFO [polismath.components.postgres:92] - modpoll 1661542778118
polis-math         | 22-09-05 19:40:33 ee4b3bf5bb6a INFO [polismath.components.postgres:75] - poll 1661542778121
polis-math         | 22-09-05 19:40:34 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :moderation > 1661542778118
polis-math         | 22-09-05 19:40:34 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :votes > 1661542778121
polis-math         | 22-09-05 19:40:34 ee4b3bf5bb6a INFO [polismath.components.postgres:92] - modpoll 1661542778118
polis-math         | 22-09-05 19:40:34 ee4b3bf5bb6a INFO [polismath.components.postgres:75] - poll 1661542778121
polis-math         | 22-09-05 19:40:35 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :votes > 1661542778121
polis-math         | 22-09-05 19:40:35 ee4b3bf5bb6a INFO [polismath.components.postgres:75] - poll 1661542778121
polis-math         | 22-09-05 19:40:35 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :moderation > 1661542778118
polis-math         | 22-09-05 19:40:35 ee4b3bf5bb6a INFO [polismath.components.postgres:92] - modpoll 1661542778118
polis-math         | 22-09-05 19:40:36 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :votes > 1661542778121
polis-math         | 22-09-05 19:40:36 ee4b3bf5bb6a INFO [polismath.poller:24] - Polling :moderation > 1661542778118
polis-math         | 22-09-05 19:40:36 ee4b3bf5bb6a INFO [polismath.components.postgres:75] - poll 1661542778121
polis-math         | 22-09-05 19:40:36 ee4b3bf5bb6a INFO [polismath.components.postgres:92] - modpoll 1661542778118
polis-math exited with code 0

[patcon]

It's been awhile, but I think most of those polis-math logs are expected, and just the polis-math exited with code 0 is the one that shouldn't be happening

Is polis working fine when not trying to connect to math with nREPL? You'll know if you submit fake votes for 7 anon users and the visualization shows up (incognito tabs should work, so long as you allow voting without login)