-
Notifications
You must be signed in to change notification settings - Fork 3.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
LS > 1.5.1 double ingestion of log when rotated based on glob path #3674
Comments
It appears that even a single glob in the filename (no glob in path), still triggers this condition |
To easily reproduce the problem with basic file input/output (used logstash 1.5.2):
And generating logs slowly so as to give logstash time to discover new logs (default discover_interval 15 seconds)
Data written to 1st log (that is rotated) is output-ed twice to output log.
|
This bug was introduced with jordansissel/ruby-filewatch#51 |
After merging the filewatch PR we still need to release a new version of the filewatch library and then update the .lock file for the 1.5.4 release |
Filewatch 0.6.5 has been released with a fix for this. |
logstash-input-file 1.0.1 has been released, needs to be included in the next release of logstash (1.5.4) |
manually installing logstash-input-file 1.0.1 plugin in logstash 1.5.2 appears to resolve the issue when running the test detailed in #3674 (comment). I'll run a production test to validate fix with tomcat logs rotated by tomcat at 00:00 UTC |
Thank you for testing this @thomasquatorze |
Updated the logstash-input-file plugin to 1.0.1 on a production host running logstash 1.5.3 and it properly handled log rotation by tomcat/log4j at 00:00 UTC. This issue appears to be resolved. Thanks |
Thank you again! this will be included in 1.5.4 |
Starting with LS 1.5.2 logs referenced with a glob path start being ingested a second time as the log is being rotate at 00:00 UTC. sincedb file content initially shows the log (inode 16648175) changing name (access_logstash.log -> access_logstash.log.2015-07-30) but then no longer being tracked. Kibana shows doubling of log volume starting 00:00 UTC the previous day. stopping ls, removing the sincedb file and starting ls resets the problem until the next rotation.
Testing was run on 2 identical systems. The one using no globs in path, did not experience the problem.
os: Ubuntu 12.04.5 LTS (precise)
pkg: logstash_1.5.3-1_all.deb (025e540e9276946871396a739042acbdd93d4e6f)
java: oracle 1.7.0_51
The text was updated successfully, but these errors were encountered: