Raising priority to High and targeting the bug to 9.0.
It's observed that when rsyslog service is running the performance degrades significantly, e.g. for keystone (apache log, the duration is in 8th column in microseconds):
On every request the service (keystone) writes only 2 messages to log, one on INFO level and the other on DEBUG. However the difference in duration is in two degrees of ten.
Raising priority to High and targeting the bug to 9.0.
It's observed that when rsyslog service is running the performance degrades significantly, e.g. for keystone (apache log, the duration is in 8th column in microseconds):
rsyslog is running: 2016:12: 56:13 +0000] "GET /v3 HTTP/1.1" 200 247 777912 "-" "curl/7.35.0" 2016:12: 56:14 +0000] "GET /v3 HTTP/1.1" 200 247 318353 "-" "curl/7.35.0" 2016:12: 56:15 +0000] "GET /v3 HTTP/1.1" 200 247 303048 "-" "curl/7.35.0"
- - - [13/May/
- - - [13/May/
- - - [13/May/
rsyslog is stopped: 2016:12: 56:20 +0000] "GET /v3 HTTP/1.1" 200 247 8799 "-" "curl/7.35.0" 2016:12: 56:21 +0000] "GET /v3 HTTP/1.1" 200 247 7909 "-" "curl/7.35.0" 2016:12: 56:22 +0000] "GET /v3 HTTP/1.1" 200 247 4765 "-" "curl/7.35.0" 2016:12: 56:22 +0000] "GET /v3 HTTP/1.1" 200 247 4665 "-" "curl/7.35.0"
- - - [13/May/
- - - [13/May/
- - - [13/May/
- - - [13/May/
On every request the service (keystone) writes only 2 messages to log, one on INFO level and the other on DEBUG. However the difference in duration is in two degrees of ten.