We haven't changed anything on filemaker server. Try the sync again, or move to an area with better wireless coverage. "FileMaker could not establish a connction to the MirrorSync server. But, when we release sync, we received below message : During the weekend, our VPN connection crashed and we finally managed to make it live yesterday. I set up MirrorSync in December 2020 and until last friday, everything was running very smoothly. Will look into it.I don't know if a similar topic has already been opened and discussed by I'm facing a kind of major issue and can't solve it. Thanks for the reply and suggestions! After researching it more it looks like the heroku hobby tier I am on only allows for 20 connections max so that may be too few for the setup I have (next.js + prisma). Prisma pool still being buggy since the connection reaper is not done properly (I reported the problem a long time ago) so the engine keeps more connections then necessary Too many concurrent instances of the service runningĮach instance holding too many connections in their pool (configured in prisma with the connection params or defaults to NUMBER CORES + 1) Each instance of your service will open a few connections to the database to improve concurrency thus throughput (otherwise you would only be able to do one DB request at a time, even if you server had hundreds of concurrent incoming requests). This means the database system/VM has a maximum number of concurrent connections it can hold and its materialized by a setting ( max_connections in postgres). In postgres and other traditional DBMS, each connection takes either a thread or a process thus consuming quite a bit of memory. Hope this means you have exhausted the number of connections that your database can accept. The alternative is to use a connection pooler like pgbouncer for postgres, but that comes with a lot of caveats in prisma (it is not super well supported from my experience).
#Mirrorsync error too many client records update#
My suggestion is usually to modify your database setting to increase the max number of connections (check current level with SHOW max_connections in postgres) to 2x your normal level (so if you have 100, you set it to 200) so that the rolling update can work. Those issues are particularly acute when doing rolling updates in production as the number of instances is 2x for a short period of time. Prisma pool still being buggy since the connection reaper is not done properly (I reported the problem a long time ago) so the engine keeps more connections then necessary.Each instance holding too many connections in their pool (configured in prisma with the connection params or defaults to NUMBER CORES + 1).Too many concurrent instances of the service running.Thus in the following cases, it is possible to "exhaust" the database connections (this is easier to do with postgres since the typical number of connection is around 100): See this comment: #1983 means you have exhausted the number of connections that your database can accept. So the solution is to kindof "cache" the client in a global variable. But hot reload results in multiple initializations of this PrismaClient Usually you initialize the PrismaClient once in your application (as a singleton).
This issue happens because many platforms as nextjs (and probably nestjs as well) do hot reload of parts of your code.
#Mirrorsync error too many client records code#
it probably needs some code changes that result in a rebuilding of this file /api/graphql.ts or its imports.create a file pages/api/graphql.ts inside that, use prisma client.
0 kommentar(er)
