If I serially make Prisma calls, it takes forever ...
# orm-help
n
If I serially make Prisma calls, it takes forever to save the data. Forever as in days. It ran overnight and couldn't save all the data. If I allow parallel calls, I get errors. Anyone else experiencing this?
h
"serialize prisma calls"?
n
sorry if I call them serially. As in one-at-a-time
edited original message
h
and what do you mean by "save the data"?
what are you trying to do?
download everything stored?
n
@harmony - we are trying to make many calls in parallel and when we do Prisma errors out
h
oh
n
as a result we have to make requests sequentially which means we end up holding onto our data as we make changes to it and then finally save it at the end
h
you might be running into the free version limitations
n
we are on paid
h
hmm
n
yeah
we've had to refactor a few processes to accomodate
h
very weird
n
and it slows our system down
and also causes processes to drop if they are active during a deployment
which is not good
n
what kind of errors do you receive?
m
@noahdavis what kind of volume of calls are you looking at? I have a data generator which makes calls to my app to create new data records at a rate of about 20 in a few seconds and it works fine. Each api call saves about 15 data points
n
@max 80-150/second
h
what are you querying?
are you even using graphql?
n
are you even using GraphQL?
What do you mean? cc: @harmony
h
it just kinda surprises me how many requests you send
n
We fetch data from millions of web pages and do many complex calculations on the extracted data, we then send the relevant data to our GQL server to mutate and send to our UI specific data store which is what we are using Prisma for.
We also use our GQL server as a proxy to@our other backend services and execute a lot of business logic on the server as well
We have been having a lot of success with our set up but have been running into blocking issues with too many parallel requests, even with throttling.
m
what are your prisma logs showing? When you're using your mutations are you connecting them to existing nodes or is the data all standalone?