Israel Fruchter

> The problem is important enough we'll analyze and solve it. Just pushing it to more versions > is useless. First QA need to analyze this, it's not enough to...

Well the disconnect backtraces are normal, and we see them in c-s when nodes go down.

> Looking at the scylla.yaml, there's `commitlog_sync: periodic` and `commitlog_sync_period_in_ms: 10000`. AFAIU this means that we save the commitlog to disk every 10 seconds. > > But what happens if...

why not ? cause there's no exceptions and no classes ?

Later I've found out the ParamikoMachine, but not much luck with getting it to work on Windows. together with rpyc ZeroDeply feature, I was just hanging there when calling `server.classic_connect()`...

> Maybe using sendall is much simpler. https://github.com/paramiko/paramiko/blob/main/paramiko/channel.py#L825 send a PR with such a change, and if all tests would be pass, we'll merge and release a new version

@scylladb/seastar-maint can someone give a hand with this one ? the only approach I've been able to conjure here is to duplicate the handlers.

sharing one use case of using `--prefix` and `--root`, inside a multi stages docker build, one stage is building/collecting all packages and install them into a /build directory and other...

@charliermarsh I don't think the --python solved the equivalent `pip install --root` case, you want us to open a new issue for that ?

history show this issue with test_simple_kill_node_while_decommissioning, is happening since September: https://70f106c98484448dbc4705050eb3f7e9.us-east-1.aws.found.io:9243/goto/6a77bc60-c5d2-11ee-81c7-3986d18dafd5 test_simple_decommission_node_while_query_info was refactor recently in https://github.com/scylladb/scylla-dtest/pull/3814 by @bhalevy, so it's probably something new as part of that.