There could be many reasons:
But my final solution used the retry option in replication ( https://pouchdb.com/api.html#replication ), forcing me to retry in the event of a failure
PouchDB.replicate(localDB, baseRemoteDB, { live: true, retry: true, timeout: false } )
As a result, replication is suspended every two documents (instead of stopping with an unknown error), and it resumes after a few moments, starting from the first new document. Response time can also be set using the snooze function (see previous document)
source share