- gawk
- gnuplot
For convenience, you may want to add this repo to $PATH instead of using absolute paths.
Below is a brief description of each job.
Summarize the count, average duration, and 99th percentile duration of per endpoint in Puppet server and DB access logs per 30 minute interval. Can also graph the count and average duration in gnuplot
top-api-calls.sh [-b] [-g] [-c] <access_files>
-b uses the JRuby borrow time instead of overall duration
-g plots the average durations in gnuplot
-c can be used in combination with -g to plot counts
herd_plot.sh <herd_query_file>
Plots the results of the thundering herd query in gnuplot. TODO: summarize the data so it's more legible
partition_sum.sh [ -t <partition name> ] <path_to_db_relation_sizes_by_table.txt>
Outputs the total size of a partitioned postgres table, defaulting to resource_events. Currently, -t pe-puppetdb.reports is also supported.
$ partition_sum.sh -t pe-puppetdb.reports resources/db_relation_sizes_by_table.txt | numfmt --to=si
29G
puppet_access_slowest.awk <path_to_access_log>
Prepend the desired field from the access logs (currently, %b for byte size of the payload and %D for the duration) to each line. Can be used in combination with other tools, e.g.
puppet_access_slowest.awk logs/puppetserver/puppetserver-access.log | sort -nr | cut -f2- -d ' ' | head
puppetdb_longest_queries.awk <path_to_puppetdb.log>
Calculate the average duration of a PDB command per node. Can be used in combination with other tools, e.g.
echo "Average Count Hostname" && puppetdb_longest_queries.awk logs/puppetdb/puppetdb.log | sort -rn | head | column -t
Calculates the average catalog compilation time per node. Can be used in combination with other tools, e.g.
echo "Average Count Hostname" && puppetserver_longest_compiles.awk logs/puppetserver/puppetserver.log | sort -rn | head | column -t