When we enter "pipeline" mode, the current connection is changed to a pipeline.
There is no problem with that, it allows to handle all get via the pipeline without having to think about it.
The database is also shared between models, for the same exact reason.
So they share the same connection to redis.
And because of the connectionpool, the same connection is always used, even with different threads.
So in the current thread we are aware that we are in pipeline mode, because we triggered it, but in the other threads, we are not, and when they try to get data, they get a pipeline in return, not the expected data.
I currently have no idea on how to resolve this, but this is a real problem.
twidi
commented
7 years ago
When we enter "pipeline" mode, the current connection is changed to a pipeline. There is no problem with that, it allows to handle all get via the pipeline without having to think about it.
The database is also shared between models, for the same exact reason.
So they share the same connection to redis.
And because of the connectionpool, the same connection is always used, even with different threads.
So in the current thread we are aware that we are in pipeline mode, because we triggered it, but in the other threads, we are not, and when they try to get data, they get a pipeline in return, not the expected data.
I currently have no idea on how to resolve this, but this is a real problem.