[chargebee] webhook failure when updating expired credit card

[JanKoehnlein]

Bug description

Chargebee reported a webhook fauilure on an "Payment Source Expiring" event.

Our payment-endpoint logs

"QueryFailedError: ER_DUP_ENTRY: Duplicate entry '<payment ref redacted>-<resourceVersion redacted> for key 'PRIMARY'
    at QueryFailedError.TypeORMError [as constructor] (/app/node_modules/typeorm/error/TypeORMError.js:9:28)
    at new QueryFailedError (/app/node_modules/typeorm/error/QueryFailedError.js:13:28)
    at Query.<anonymous> (/app/node_modules/typeorm/driver/mysql/MysqlQueryRunner.js:222:57)
    at Query.<anonymous> (/app/node_modules/mysql/lib/Connection.js:526:10)
    at Query._callback (/app/node_modules/mysql/lib/Connection.js:488:16)
    at Query.Sequence.end (/app/node_modules/mysql/lib/protocol/sequences/Sequence.js:83:24)
    at Query.ErrorPacket (/app/node_modules/mysql/lib/protocol/sequences/Query.js:92:8)
    at Protocol._parsePacket (/app/node_modules/mysql/lib/protocol/Protocol.js:291:23)
    at Parser._parsePacket (/app/node_modules/mysql/lib/protocol/Parser.js:433:10)
    at Parser.write (/app/node_modules/mysql/lib/protocol/Parser.js:43:10)"

Steps to reproduce

It looks like this was the first time a user used a credit card in the same month in which it was expiring.

Workspace affected

No response

Expected behavior

No response

Example repository

No response

Anything else?

No response

[JanKoehnlein]

We use paymentSource (VARCHAR) and resourceVersion (BIGINT) as primary keys in the d_b_paymnet_source_info. When starting a subscription with an expiring creditcard, Chargebee sends events of type "payment source added" and "payment source expiring" with the same resourceVersion (usually this is icremented between events).

Looks like TypeORM doesn't correctly deal with primary keys of number types, so it fails to find the current entry but fails to add a new one on save.

[JanKoehnlein]

Maybe related to typeorm/typeorm#6060 or typeorm/typeorm#2930

[JanKoehnlein]

/schedule

[jankeromnes]

@JanKoehnlein Random after-thought: Wouldn't repo.update(primaryKey, shape) have resolved the problem of duplicate entry that repo.save(shape) has?