How do I make Prisma subscriptions more reliable?

In production, they are very temperamental...

Hi! Is anyone running into an issue where Prisma Studio can’t open because of this error:

This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). The promise rejected with the reason:

When I first launch it I have no issues, but after I make a modification to the DB by creating a new entry for example, I see this issue and have to restart Prisma Studio to see my data again. Here is the rest of the error message (I am deploying this as a AWS service):

TypeError: Converting circular structure to JSON

--> starting at object with constructor 'Object'

|     property 'fields' -> object with constructor 'Array'

|     index 0 -> object with constructor 'Object'

|     ...

|     index 0 -> object with constructor 'Object'

--- property 'type' closes the circle

at JSON.stringify (<anonymous>)

at serialize (/usr/local/lib/node_modules/prisma/build/index.js:131814:17)

at HTTPNodeTransport.respond (/usr/local/lib/node_modules/prisma/build/index.js:131829:34)

at /usr/local/lib/node_modules/prisma/build/index.js:140730:29

at processTicksAndRejections (internal/process/task_queues.js:97:5)

Has anyone encountered an issue where the prisma 1 docker container will randomly hang up queries with ECONNRESET? It's been happening on our production servers for the last week and has been driving me mad!

Possibly related to my last post, I've found that some queries have been causing prisma to die and restart within the container:

/app/start.sh: line 4:     7 Killed                  /app/bin/prisma-local
No log level set, defaulting to INFO.
[INFO] {} - Starting...
[INFO] {} - Start completed.
[INFO] {} - Starting...
[INFO] {} - Start completed.
[INFO] {} - Shutdown initiated...
[INFO] {} - Shutdown completed.
[INFO] Initializing workers...
[INFO] Obtaining exclusive agent lock...
[INFO] Successfully started 1 workers.
[INFO] Obtaining exclusive agent lock... Successful.
[INFO] Deployment worker initialization complete.
Server running on :4466
[INFO] {} - Starting...
[INFO] {} - Start completed.

The query is filtering rows from a table with 80 columns and ~45k rows, the widest table in the application but by no means the longest.

There's nothing more that the above in the container logs - they're entirely normal until the process is killed out of the blue

Are there any other logs in the container that might have more information?

@Moray Macdonald Are you able to see the memory use when this happens?

Running out of memory was my first thought but I've not been able see it yet. I can't reproduce the issue reliably and as I can't forward stats from inside the container to Zabbix for later analysis I'm stuck watching

docker stats

and hoping I catch it! The container has 8GB available and normally usage is 1.5-2GB, so something must be very wrong if it is running out!

Is there any french dev familiar with prisma in here interesting in a mission of a few days ? thank you

Good morning there! I've been running into some problems when using the docker image prismagraphql/prisma So currently we're using prisma1 at work because upgrading to prisma2 is planned but not done yet. I've bought a m1 macbook but prismagraphql/prisma won't work. I've rosetta installed and done the few things someone find on the internet, but not working. When I run docker-compose it starts but then exit with the error message:

ds_prisma   | qemu: uncaught target signal 11 (Segmentation fault) - core dumped
ds_prisma   | /app/start.sh: line 4:    14 Segmentation fault      /app/bin/prisma-local
ds_prisma exited with code 139

I don't know if there is a possibility to self recompile the image on another machine to run on arm64 or what is the problem, but it would be sad if this is the only thing which stops me using this laptop. Thanks in advance for every help 🙂

Hi, does anyone know if it’s possible to run multiple prisma instances against the same mysql database safely?

Do anyone know if there are prisma-to-UML (or other diagrams) that work for Prisma 1-style of schema?

Hi! How to fix an error? P.S. macbook m1

HELP. Our production prisma containers are now no longer responding to any requests of any kind. I've tried rebuilding the containers and re running

prisma deploy

, neither of which makes a difference. Nothing has changed in our hosting/network configuration. This has taken our application down completely and no indication of why. HELP. https://prisma.slack.com/archives/C0152UA4DH9/p1626795907030200 https://prisma.slack.com/archives/C0152UA4DH9/p1626959263035200

Hi , not sure if this is the right place but we use Prisma 1 and looking to rename some of the tables. Is there any easy way to achieve that via Prisma at all?

I’ve been posting to the Prisma 2 channel apparently. I just saw a pop up saying there was a Prisma1 channel: I have quite a few issues posted on the #general channel that were supposed to be here. createdAt and updatedAt issues: https://prisma.slack.com/archives/CA491RJH0/p1629813591186100 prisma1 deploy migration breaks relations because it deletes relation table ids then says they are missing: https://prisma.slack.com/archives/CA491RJH0/p1629406021068400 prisma1 introspect does nothing after completing wizard with all correct information: https://prisma.slack.com/archives/CA491RJH0/p1629812052182500 Any help would be greatly appreciated!

Is there any way to prevent the new data model v1.1 from removing relation table ids? It tries to remove them then complains that they are not there and fails and it is preventing deployment of the rest of my data model.

prisma1 export fails to export anything with no error:

prisma# prisma1 export
Warning: The `prisma1 export` command will not be further developed in the future. Please use the nativeexport features of your database for these workflows.

More info here:
MySQL: <https://dev.mysql.com/doc/refman/5.7/en/mysqlimport.html>
Postgres: <https://www.postgresql.org/docs/10/app-pgrestore.html>

Downloading nodes 2028ms
Downloading lists 45ms
Downloading relations !
 ▸    [object Object]

Get in touch if you need help: <https://slack.prisma.io>
To get more detailed output, run $ export DEBUG="*"
(node:1773) [DEP0066] DeprecationWarning: OutgoingMessage.prototype._headers is deprecated
(Use `node --trace-deprecation ...` to show where the warning was created)
prisma# ls
datamodel.graphql  enums.graphql  prisma.yml  relations.graphql  types.graphql

I am trying to use some fragments serverside, but how can I pass parameters/filters? I have { id, processes { id, state }, etc } but where I'd like to only get the processes that have the state "ACTIVE"

Is the export path command for prisma 1

prisma export --path

or

prisma export --export-path

for 1.30.5? The documentation conflicts with itself.

I just ran a pg_dump on my postgres database of 2.8GB and it took about 15-20 minutes to complete. prisma export (v1.30.5) takes 11+ hours and it is still not done. Any ideas how and why prisma export would take so long compared to pg_dump? I need prisma export/import because it is the only way I have been able to successfully upgrade my datamodel to v1.1 and then eventually to version 2. However, I can’t have a full day of downtime just to back up my database…

@inlightmedia We still use Prisma 1 too, but v1.30.5 seems even older? We have 1.34.10

@Lars Ivar Igesund We used to use an even earlier version 1.13, and we kind of struggled to upgrade to Prisma 2 since we skipped all later versions, directly to 2

Hi! With many switching to Prisma 2 or 3, how likely is it that support will end here at Slack for Prisma1 as well? At Github the support is at a minimum, is this what will happen here too? Are there deadlines for when companies should have migrated to 2 or 3?🙂

So I have noticed that new IDs in prisma1 come in sequence if generated in close proximity (in terms of time) to eachother, e.g. 6151.......1a8 is followed by 6151........1a9. Is there some way to make them less serialized?

Another question - this is about getList operations (we have a graphql API generated by old nexus stuff). When I do a get list with a search filter, and e.g. get 1 matching result, the count property is set to the amount of objects available in the database (mongo), so usually a larger number. I would expect the count/total to match the actual results to be returned? I know there are some aggregate issues (for each collection, I need to manually add an aggregate field to the connection types), but perhaps the implementation is too simple for the setup. Now this field returns root.aggregate in the resolver, so maybe that is too simple - are there any good examples/suggestions on how to get this to work?

Hi, I'm setting up my application stack on a new dev machine with prisma 1 running in docker. Whenever I try to start it, I get the following error:

qemu: uncaught target signal 11 (Segmentation fault) - core dumped
/app/start.sh: line 4: 13 Segmentation fault /app/bin/prisma-local
prisma exited with code 139

I'm running on a Mac with an M1 processor if that makes a difference. I've tried adding

platform: linux/x86_64

to the compose file, which hasn't done anything

I'm guessing this is an Apple Silicon problem. Are there any instructions for recompiling the Docker image? I can't find anything in the docs or on Github. I'm not even sure what image the Docker image is built on!

Looking to hire Prisma1 expert for scaling & performance issues. Please feel free to reply or DM me.