Due to ArangoDB's lack of document-level locks, any DB operation that could encounter write conflicts must be rigged to retry the transaction as appropriate. See the data/create method for an example. Note that this is only for allowed conflicts, such as shared metrics updated, real conflicts (two processes modifying a doc in different ways) should be treated as errors.
dvstans
commented
3 years ago
Due to ArangoDB's lack of document-level locks, any DB operation that could encounter write conflicts must be rigged to retry the transaction as appropriate. See the data/create method for an example. Note that this is only for allowed conflicts, such as shared metrics updated, real conflicts (two processes modifying a doc in different ways) should be treated as errors.