After learning a bit more about parsing using grok I've found a lot of the time it isn't necessary to have to write my own regexes. There are a number of predefined grok patterns I can use and I can extend these to create my own custom patterns when parsing logstash logs.
A useful link on the grok patterns supported by logstash: https://github.com/elastic/logstash/blob/v1.4.2/patterns/grok-patterns.
Using the new found knowledge I was able to change my match configuration to that below.
if[type] == "mytype" {
grok {
match => ["\[%{TIMESTAMP_ISO8601:dateTime}\]%{SPACE}\[%{WORD}\:%{FLOATINGPOINT:cpu}\]%{SPACE}\[%{WORD}\:%{FLOATINGPOINT:memory}\]"]
}
}
This uses the built in grok patterns TIMESTAMP:ISO8601 to pick out the date in my logs, and I have created a very simple custom pattern FLOATINGPOINT to pick out the floating point values for memory and cpu in my example. The FLOATINGPOINT pattern looks like:
FLOATINGPOINT %{INT}\.%{INT}