[GOWS] - Memory leak with a lot of /messages requests

[matteus-siqueira]

I migrated a provider with 150 clients from NOWEB to GOWS and unfortunately I had the experience of GOWS breaking down with many requests because it was a chatbot serving 24/7. I had to return to NOWEB, the server consumption using GOWS was absurdly higher than with NOWEB.

[devlikepro]

for the history - we found that it's likely because of frequent GET chat/messages?limit=1 requests. The memory leak happens likely there's a memory leak in grpc server on golang side. Need to check it and see what we can do with such load.

Because yeah, it's not because of the number of sessions, here's another installation:

[matteus-siqueira]

Yes, unfortunately the API did not support a high load as noweb does.

[mequelussi]

for the history - we found that it's likely because of frequent GET chat/messages?limit=1 requests. The memory leak happens likely there's a memory leak in grpc server on golang side. Need to check it and see what we can do with such load.

Because yeah, it's not because of the number of sessions, here's another installation:

What's the secret to making this work?

Could you share the confs and the structure you set up for this?

[devlikepro]

@mequelussi just regular GOWS, no specific options required

https://waha.devlike.pro/docs/how-to/install/

[mequelussi]

Are you using postgres or local storage?

[devlikepro]

@mequelussi the instance uses local storage (sqlite3 db basically)

[devlikepro]

Hi! Found why it happened, starting from 2025.6.1 there shouldn't exist that memory leak (but may exist another 😆 ) 🤞