timeseries_delete_with_meta_concurrent.js and timeseries_update_concurrent.js have not been running against sharded clusters since they are marked as assumes_against_mongod_not_mongos because they use the hangDuringBatchRemove and hangDuringBatchUpdate failpoints which don't exist on mongos.

However, it looks like they also don't work on sharded cluster with embedded router (i.e. the test connects directly to the primary mongod on the shard). They both hang during the drop here and here. We should investigate if this is because of the test setup or if this is because the behaviors expected in the tests do not hold on sharded clusters. If the former, we should adjust the test accordingly. If the latter, we should file a ticket to address the feature parity issue.