[
https://issues.apache.org/jira/browse/LOG4J2-44?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13233705#comment-13233705
]
Tushar commented on LOG4J2-44:
------------------------------
https://issues.apache.org/jira/browse/LOG4J2-59 issue for latest build
i will show you the async appender sample tomorrow - what would be good to have
is when it gets very busy to increse its buffer or drop items OR block when TOO
big (optimal and MAX size) - all as properties
right now are prod breaks as threads get blocked when too much logging happens
while many txns are happening ... i would rather see log4j stop logging (as a
preference) then block and causing are server to crash. the only long term fix
we found was to log less (we were logging too many times at places and at one
meta data reload - logging too long a string - a whole hash map made of a whole
'static' data table)
but at other times too we have seen thread dead lock and the histo and thread
dump via jstack says there were log4j objects .... i can copy paste the exact
eroror msg if you need
> log4j java date wise log every X MB
> -----------------------------------
>
> Key: LOG4J2-44
> URL: https://issues.apache.org/jira/browse/LOG4J2-44
> Project: Log4j 2
> Issue Type: Improvement
> Components: Appenders
> Reporter: Tushar
> Priority: Minor
>
> a log roll over that makes a new file every x MB/ KB but the current file is
> archived with a full date-time--random-int-to-make- unique stamp
> so if the file is myProj_A.log after x MB is renamed to
> myProj_A_2010_09_10_02_48_02_XX34334.log (year-month-day- time ...) and so on
> Is there a pattern/ property to do this already?
> can see http://www.simonsite.org.uk/ and
> http://www.simonsite.org.uk/javadoc/org/apache/log4j/appender/TimeAndSizeRollingAppender.html
> here the code is a little mixed up (we want test and real code in seperate
> src roots) - and not sure if it works completely but it says it does. but it
> has our requirements.
> Code should write out buffer in bits instead of blocking for a long time
> emptying full buffer when buffer is full Or is that done by nio ?
> build and run script to test with params to control logging, number of
> threads, amount of data logged so we can see that the logs are rolling as
> planned we want the final roller to roll every day but can keep that as hour
> and minute params (so final will be 24 hours and 0 minutes, but for testing
> can be 5 minutes only, similarly size can be in KB) also need format of file
> name with prefix for folder so the root log folder can be c:\log then we
> specify the dynamic sub folder like yyyy\mm\dd meaning make sub folders for
> current year, inside that for month and inside that for day and finally the
> file name format with a prefix like myApp_{yyyy_mm_dd_HH_II_SS) and post fix
> so the final file will be in folder (example)
> C:\log\2010\10\29\myApp_2010_10_29_14_34_22_randXYZ.txt where randXYZ stand
> for some random text that is added to make the file name unique (as required
> or can use java inbuilt getTempFile function ...)
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org
For additional commands, e-mail: log4j-dev-h...@logging.apache.org
