A query analyzer that parses Solr's log file to get some basic query statistics
Note that you need to enable logging at the INFO level for this to work and you need to have cores enabled. You also need python3 installed.
# reading from stdin
cat file1 file2 | ./solr_loganalyzer.py --max 20
tail -1000 file | ./solr_loganalyzer.py --max 20
# create statistics and a siege urls file by continously reading the solr.log
tail -F solr.log | ./solr_loganalyzer.py --max 20 --write_file urls.txt http://127.0.0.1:10080/solr4/ | tee stats.txt
# perform a siege loadtest by using the urs file
apt-get install siege
siege -f urls.txt --concurrent=5 --reps=5 --delay=0.5 --no-parser
# reading one or more files
./solr_loganalyzer.py solr.log solr.log.1 solr.log.2
The analyzer outputs statistics grouped by Solr core. Here is an example:
Top Endpoints for core1
========================================
1) "/mlt" 3 times
Top Searh URLs for core1
========================================
1) "mlt.count=16&&q=qrows=16" 1 times
2) "mlt.count=16&start=0&q=&rows=16" 1 times
Slowest Searches for core1
========================================
1) "mlt.count=16&&q=qrows=16" 30 times
2) "mlt.count=16&start=0&q=&rows=16" 11 times
Top Factet Fields for core1
========================================
ITEM 1 : 3384 times "language"
ITEM 2 : 1578 times "authorName"
Search Time for core1
========================================
Median 2
75% 11
90% 11
99% 30
nostests -s
This tool is inspired by redis-faina. The tool was forked from solr-loganalyzer.
This tool is distributed with the BSD-2 Licence.
- Didier Deshommes dfdeshom@gmail.com
- Marc Schoechlin marc.schoechlin@vico-research.com