mk_filestats: show corret summary if running as clustered service

2 votes

Running mk_filestats as clustered service displays an incorrect summary because the summary is taken of the best/worst node, which does not represent the metrics of the entire cluster.

Best: [host1], Files in total: 1, Smallest: 344 B, Largest: 344 B, Newest: 1 minute, Oldest: 1 minute
Worst: [host2], Files in total: 1, Smallest: 894 B, Largest: 894 B, Newest: 1 year, Oldest: 1 year
Expected: Files in total: 2, Smallest: 344 B, Largest: 894 B, Newest: 1 minute, Oldest: 1 year

On a multi node cluster the files don't have to be on the same node. mk_filestats should merge first the data of all nodes and then evaluations afterwards (as if it were a single host). This would show the metric of the smallest/largest/youngest/oldest File in the Summary regardless of which node the file is on. In failover mode, it should checked additionally whether all data only comes from the active, standby or both. In the details, the data should be displayed per node as is already the case today.

Under consideration Checks&Agents Suggested by: Lars Sörensen Upvoted: 26 Mar Comments: 0

Comments: 0