Rev | Line | |
---|
[154] | 1 | -- This script processes failure cause for different categories of events grouped by duration (short, medium, long) |
---|
| 2 | |
---|
| 3 | -- author: cristina |
---|
| 4 | |
---|
| 5 | REGISTER failurecausefunc.jar; |
---|
| 6 | |
---|
| 7 | raw = LOAD '$inputDir/event_trace.tab' USING PigStorage('\t') AS (event_id,component_id,node_id,platform_id,node_name,event_type,event_start_time,event_stop_time,event_end_reason:chararray); |
---|
| 8 | |
---|
| 9 | -- build durations |
---|
| 10 | duration = FOREACH raw GENERATE event_stop_time-event_start_time as dur, event_end_reason as fault; |
---|
| 11 | mapping = FOREACH duration GENERATE flatten(RealLabels(dur, fault)) as key, 1 as value; |
---|
| 12 | --filteredMapping = FILTER mapping BY key!=NULL; |
---|
| 13 | |
---|
| 14 | |
---|
| 15 | -- group and average |
---|
| 16 | grup = GROUP mapping BY (chararray)key; |
---|
| 17 | rez = FOREACH grup GENERATE group, SUM(mapping.value) as sum; |
---|
| 18 | filteredRez = FILTER rez BY sum > 1000; |
---|
| 19 | |
---|
| 20 | STORE filteredRez INTO '$outputDir' USING PigStorage(); |
---|
Note: See
TracBrowser
for help on using the repository browser.