I was building the URI in a spring configuration file as
spring.data.mongodb.uri=${MONGODB_URI}/test?retryWrites=true&w=majority

where MONGODB_URI came from a kubernetes secret (as it contains user & password)

somehow a new-line had slipped into the kubernetes secret, so the query was effectively asking for "fpos-dev-pl-1.kqybg.mongodb.net\n"

Besides learning how to correctly configure secrets, I learned in the process that neither Spring nor mongo was validating that spring.data.mongodb.uri was a well-formed URI and the application failed to start with error messages relating to not finding DNS and not being able to resolve the TXT record. This was very confusing.

The driver should accept a URI parameter (instead of a String) and ensures that it is valid.

Long-running transactions may trigger an aborted transaction, but the cause is not clearly communicated by the driver exception message.

Failed: Command insert failed: Transaction 1 has been aborted..
Unhandled exception. MongoDB.Driver.MongoCommandException: Command insert failed: Transaction 1 has been aborted..

To reproduce, set a very low timeout:

replicaset:PRIMARY> db.adminCommand( { setParameter: 1, transactionLifetimeLimitSeconds: 1 } )

Then run a long running transaction from the C# driver.

The cause can be confirmed by checking the MongoDB log for a [abortExpiredTransaction]:

replicaset:PRIMARY> show log global

2020-07-31T03:59:49.196+0000 I COMMAND [abortExpiredTransactions] Aborting transaction with txnNumber 1 on session xxxxxx because it has been running for longer than 'transactionLifetimeLimitSeconds'
2020-07-31T03:59:49.196+0000 I TXN [abortExpiredTransactions] transaction parameters:{ lsid: { id: UUID("xxxxxx"), uid: BinData(0, xxxxxx) }, txnNumber: 1, autocommit: false, readConcern: { level: "snapshot" } }, readTimestamp:Timestamp(0, 0), terminationCause:aborted timeActiveMicros:51529 timeInactiveMicros:7004962 numYields:0 locks:{ ReplicationStateTransition: { acquireCount: { w: 3 } }, Global: { acquireCount: { w: 2 } }, Database: { acquireCount: { w: 2 } }, Collection: { acquireCount: { w: 2 } }, Mutex: { acquireCount: { r: 2 } } } storage:{} wasPrepared:0, 7056ms

It would be extremely useful for tracking down the cause if the driver exception message could communicate more detail as contained in the server log, e.g.,

"Aborting transaction with txnNumber 1 on session xxxxxx because it has been running for longer than 'transactionLifetimeLimitSeconds'"

python and other drivers offer to-json utils. This makes taking a native rich shape complete with datetimes and byte[] and such and easily turning into bson (the best way!) or JSON. The utils offer options to modify the output representations of types but always do so in an EJSON way, namely with $date/$numberDecimal etc. Sometimes a consumer cannot (or will not) accept data in this fashion. I'd like to see a "safePureJSON" option (or similar) for bson.json_utils.dumps() that emits the safe string or number equivalent of the EJSON.
fld: {$date: "ISOdate"} becomes fld: "ISOdate"
fld: {$numberDecimal: "99.9"} becomes fld: "99.9" (string, to prevent float parse issues)
fld: {$numberLong: "1234567890123456"} becomes fld: 1234567890123456

float and int32 are not a problem.
Not as much of a issue for $binary because the extra info is probably helpful for the decoding rather than a hindrance.

This is probably not Node Specific a general problem with connecting via a seed list.

ServerSelection via seed list is a great concept. It works poorly. I have attached an hour of starting a working webserver app and 30 times starting it in a little over an hour. It failed to connect over 40 % of the time. A larger amount of successful connections were made over 35 seconds which means they would fail in a hosting environment like Heroku. Only 10% of the time did it connect in < 6 seconds. Your support team is avoiding using seed lists in their troubleshooting. A tool would allow them to get the data of the API calls in question Lastly, this would cleanup a lot of bad answers to connectivity issues and then the port driver team could fix the problem.