BF-36751 is an example for the concurrency_sharded_local_read_write_multi_stmt_txn_with_balancer task hitting the Evergreen timeout.

Many of the tests in that task concurrently write to the same documents, so they can create a lot of write conflict errors and retries.
The total task duration can vary greatly depending on how operations overlap, and also the state of the balancer that is enabled in the suite and that performs random resharding.

BF-36751 contains an example of a test from that suite taking ~9 minutes, and other examples in which the same test took ~30 seconds.

The total task duration sometimes gets slightly above 120 minutes, and then the task fails. This makes me believe that we should slightly increase the task timeout.