Details
-
Improvement
-
Resolution: Unresolved
-
Major
-
2.5.1
-
Security Level: Public
-
Large number of nodes.
Description
ns_server.stats.log contains the statuses of every node in the cluster from ns_doctor:
[ns_doctor:debug,2014-07-10T0:58:24.510,ns_1@cb-01.lan:ns_doctor<0.3026.0>:ns_doctor:handle_info:167]Current node statuses:
As the number of node increases the ns_doctor output gets bigger and as a result we lose the historical data that mortimer uses from stats_collector:
[stats:debug,2014-07-10T5:25:03.967,ns_1@cb-01.lan:<0.5614.0>:stats_collector:log_stats:136](at 2014,7,10},{5,25,3 (1404984303962)) Stats for bucket "Default":
This means on large cluster we only get a few hours of data that mortimer can use. We also have this problem when there is a lot of buckets.
It might be worth putting the doctor information into a different file: ns_server.doctor.log and maybe a stats file per a bucket: ns_server.stats-<BUCKET NAME>.log.
It would be good to get input from other people in the field team.