[
https://issues.apache.org/jira/browse/MINIFI-296?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16023610#comment-16023610
]
ASF GitHub Bot commented on MINIFI-296:
---------------------------------------
Github user phrocker commented on the issue:
https://github.com/apache/nifi-minifi-cpp/pull/104
This is good stuff. I'm taking a look
General comments before I dive in. Did we lose the ability to change the
logger format real time? If there was an issue in production and we had to
change the logger to a null appender, restarting may not always be possible,
whereas sending a shared object through a p2p c2 protocol would allow us to
dynamically load and change loggers without impacting the implementation, for
example ( kind of a contrived example)
Additionally, I Noticed that my logs rolled much more frequently because we
have an additional 52 bytes per line in some cases. Is there a way to limit
that? It slowed my runs down because there was much more memory allocated by
spdlog log statements when threading was high and I tested with trace. Even to
the class name would be useful.
> More configurable logging
> -------------------------
>
> Key: MINIFI-296
> URL: https://issues.apache.org/jira/browse/MINIFI-296
> Project: Apache NiFi MiNiFi
> Issue Type: Improvement
> Components: C++
> Reporter: marco polo
> Assignee: Bryan Rosander
> Priority: Minor
>
> The logging functionality would be more useful if it could be tuned on a
> per-class basis. This would allow us to set more detailed log levels for a
> place where trouble is suspected while reducing noise from other areas.
> Composable log appenders would allow us to have multiple sinks so that we
> would have a log appender that could "phone home" with information while
> concurrently logging to disk. Using a processor to do the same may be too
> onerous; however, it stands to reason that we may use the processor as the
> delivery mechanism, so we may eventually negate this issue entirely if it is
> decided that we should ship the log itself via a processor.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
