In the case of accessing an unlimited collection by using this adapter and other methods (such as the Data Explorer in Azure portal) together, limitations are as follows.
Each operation of this adapter operates on the following premises.
Scope of ID uniqueness: ID is unique in the collection.
Partition key: The field storing partition key exists in the document.
Operations are not supported if you perform any of the following actions by the methods other than this adapter.
Create documents of the same ID in the multiple partitions
Number type in Azure Cosmos DB is IEEE754 double precision.
If the number that exceeds IEEE754 double precision is written to Azure Cosmos DB as BigDecimal type, it will be rounded off. (It will not be an error)
To avoid rounding off, set schema type as String type when writing the document.
However, as it is saved as string type in Azure Cosmos DB, SQL for number will no longer be able to be used.
"_rid", "_ts", "_self", "_etag", "_attachments", etc. are the items generated by Azure Cosmos DB for a document.
You cannot write arbitrary value for these items.
As name of item generated by Azure Cosmos DB starts with an underscore (_), it is recommended to set item names for schema such that there is no underscore at the start of the names.
For more details on items generated by Azure Cosmos DB system, refer to the following pages.
"System vs. user defined resources" - (https://docs.microsoft.com/en-us/azure/cosmos-db/documentdb-resources#system-vs-user-defined-resources)
In Azure Cosmos DB adapter, among the 5 types of consistency ("Strong", "Bounded-Staleness", "Session", "Consistent Prefix", "Eventual"), "Session" is used.
For more details on consistency of Azure Cosmos DB, refer to the following page.
"Tunable data consistency levels in Azure Cosmos DB" - (https://docs.microsoft.com/en-us/azure/cosmos-db/consistency-levels)