Skip to content

Tags: bhalevy/scylla-ccm

Tags

scylla_node-watch_rest_for_alive-normal-token-owners-v4

scylla_node: watch_rest_for_alive: add wait_normal_token_owner param

For backward compatibility.
Some tests may want to pass `node.start(wait_other_notice=True)`
and not wait for nodes to become normal token owners
if they need to examine the node earlier than that.

Signed-off-by: Benny Halevy <[email protected]>

scylla_node-watch_rest_for_alive-normal-token-owners-v3

scylla_node: watch_rest_for_alive: add wait_normal_token_owner param

For backward compatibility.
Some tests may want to pass `node.start(wait_other_notice=True)`
and not wait for nodes to become normal token owners
if they need to examine the node earlier than that.

Signed-off-by: Benny Halevy <[email protected]>

scylla-jmx-listen-address-v2

scylla_jmx: listen on the node address:7001

Currently all scylla-jmx instances listen on a broadcast
address with a unique port number calculated as
7000 + node.id * 100 + cluster.id.
But, this port can be too high and clash with other port
that are in use if the simulated cluster has too many nodes.

Instead, restrict the instance to listen only on the node address
and a pre-defined port (same as cassandra's: 7199).

See https://cassandra.apache.org/doc/latest/cassandra/faq/#what-ports
```
By default, Cassandra uses 7000 for cluster communication (7001 if SSL is enabled), 9042 for native protocol clients, and 7199 for JMX.
```

Fixes scylladb#390

Signed-off-by: Benny Halevy <[email protected]>

refactor-start-stop-v2

scylla_node: _update_jmx_pid: print Timed out message only if wait=True

We get benign Timed out messages when stopping a node before
scylla-jmx started running and we explicitly pass wait=False
for this reason.

Signed-off-by: Benny Halevy <[email protected]>