-
Bug
-
Resolution: Done
-
Normal
-
Logging 5.5.6, Logging 5.6.5
-
False
-
None
-
False
-
NEW
-
VERIFIED
-
Parsed log messages will now also include "structured.time", if the output destination supports it. This fix adds 'keep_time_key' to the fluentd parsing config, when "parse: json" is enabled on the pipeline.
-
Bug Fix
-
-
-
Log Collection - Sprint 236, Log Collection - Sprint 237
Description of problem:
Time field of log message does not parse as structured.time by default in fluentd (when timestamp is also included).
for following log message
{ "time": "2023-04-13 13:41:40.3958", "level": "INFO", "message": "Get OpenIdConfiguration" }only for "level" and "message", fields "structured.level" and "structured.message" is created by enabling json parsing in clusterlogforwarder configuration, but the "structured.time" filed is not created in kibana.
fluentd parsing feature by default individually parses any json "time" field to the record's "timestamp", which could be verified through https://docs.fluentd.org/filter/parser#reserve_time
Version-Release number of selected component (if applicable):
RHOCP cluster version 4.11
Red Hat Openshift logging version 5.5.6 and 5.6.5
How reproducible:
can be reproduced.
Steps to Reproduce:
- Configure clusterlogforwarder and enable json parsing.
- Create index patterns with time field as timestamp.
- Check a log in kibana whose message contains time field, it wont have a structured.time key.
Actual results:
structured.time field is not present
Expected results:
structured.time field should be present
Additional info:
Customer's environment is Red Hat Openshift logging version 5.5.6 but I also found similar issue in Red Hat Openshift logging version 5.6.5 during my reproduce
The workaround for this issue is using vector.
- links to
- mentioned on