Parsing connection port fails whit whitespace after port number

[twdkeule]

When parsing the connection port, the config name/option is included because the sed rege does not match:

# grep -e '^native_transport_port: ' "$CONF_DIR/cassandra.yaml" | sed -e 's|^[^:]*:[ ]*\([^ ]*\)$|\1|' | tr -d "'"
native_transport_port: 9042

Solution: allow whitespace after port number with \w*:

# grep -e '^native_transport_port: ' "$CONF_DIR/cassandra.yaml" | sed -e 's|^[^:]*:[ ]*\([^ ]*\)\w*$|\1|' | tr -d "'"
9042

I noticed the parsing is always this strict, eg when parsing IP adresses.

[msmygit]

@twdkeule thank you for using the diagnostics collection script and providing us feedback! Happy to received PRs as well.

I noticed that the command in question is working without any hiccups on the system where I tested this and it is correctly pulling the port details for me as shown below,

$ grep -e '^native_transport_port: ' "cassandra.yaml" |sed -e 's|^[^:]*:[ ]*\([^ ]*\)$|\1|'|tr -d "'"
9042

Alternatively, I also did run your command which also yielded the same output,

$ grep -e '^native_transport_port: ' "cassandra.yaml" | sed -e 's|^[^:]*:[ ]*\([^ ]*\)\w*$|\1|' | tr -d "'"
9042

so, I don't think we need to get this changed in here at this time.

[RomainAnselin]

Ive encountered a similar problem on the listen_address. It didnt fail the collection, but ended up being unable to parse the nodes IP and the diag collected had X folders "listen_address: " which I then had to correct manually.
This is how I ended up on this one.

For this particular scenario above, to handle white spaces or making sure we target the port info only, I would consider using regex in the sed as we know a port will be defined by a set of 1 to 5 digits. ie: sed -E 's|^[^:]*:[ ]*\([0-9]{1,5}\)$|\1|'|tr -d "'"

Would also be of interest to confirm the sed version used in both tests above (OP + response) as I've encountered issues in the past on some AMI build where sed isnt up to date