Kamil Braun

> The cause of this issue could be the same as https://github.com/scylladb/scylladb/issues/16709. Seems unlikely, because the symptoms are completely opposite --- here we have too many reconnects and in that...

@avelanarius @sylwiaszunejko one test that is failing often in CI is here https://github.com/scylladb/scylladb/issues/17353 but it's a bit different situation, it's a single-node test and we don't restart the node, but...

We have a good reproducer https://github.com/scylladb/scylladb/issues/17353#issuecomment-1970969872

Branch: https://github.com/kbr-scylla/scylladb/tree/connection-reconnect-repro compile in debug mode and run ``` ./test.py --mode debug --cpus=0,1 --jobs=32 test_native_transport --repeat 100 ``` note: this is a single-node test

Updated the branch, reducing shard number on Scylla to 1 run with ``` ./test.py --mode debug --cpus=0 --jobs=32 test_native_transport --repeat 100 ``` it reproduces even more often! So apparently, it's...

Note: The reproducer mentioned above is for a single-node case, where the node disables the CQL server and re-enables it. We bisected the regression for that single-node case to d735957e3a7b7178f4fff26f6dbae588e58cf314....

@avelanarius any chance to pick this up? If it works better with `IN` queries on some clusters but better with individual queries on others, maybe the library should offer a...

> How will the user know which value to use? Testing of their own workload against their own cluster (and its version) ? I guess so. Apparently that's what @0xF0D0...

> 🔴 CI State: FAILURE https://github.com/scylladb/scylladb/issues/17662

Replaced by https://github.com/scylladb/scylladb/pull/18820, closing