`decode_netflow9` and `decode_ipfix` rescue blocks return logger object instead of events,

[donoghuc]

The decode_netflow9 and decode_ipfix methods have rescue blocks that catch parsing errors and log a warning, but the return value of @logger.warn(...) (a LogStash::Logging::Logger object) becomes the method's return value.

The callers expect an array and call .each on the result:

decode_netflow9(flowset, record).each{|event| yield(event)}

This causes:

NoMethodError: undefined method `each' for #<LogStash::Logging::Logger:0x768fa52f>

Stack trace:

[ERROR][logstash.inputs.udp] Exception in inputworker 
{"exception"=>#<NoMethodError: undefined method `each' for #<LogStash::Logging::Logger:0x768fa52f>>, 
"backtrace"=>[
"/usr/share/logstash/vendor/bundle/jruby/3.1.0/gems/logstash-codec-netflow-4.3.2/lib/logstash/codecs/netflow.rb:100:in `block in decode'",
...

The same issue exists in decode_netflow5, though it manifests differently since the caller yields directly rather than iterating.

Impact: When a BinData::ValidityError or IOError occurs while parsing a record, the NoMethodError exits the flowset.records.each loop, and all remaining records in that packet are lost.

Solution

Remove the rescue blocks from decode_netflow5, decode_netflow9, and decode_ipfix. Add error handling inside the flowset.records.each loops in the decode method instead. This ensures:

• The error is logged
• Processing continues for remaining records in the flowset

[donoghuc]

PR #214