[jira] Commented: (DERBY-5051) Update the docs to include the additional information available in derby.log. This information will be useful from product supportability stand point

["Kristian Waagan (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DERBY-5051?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13001037#comment-13001037 ] 

Kristian Waagan commented on DERBY-5051:
----------------------------------------

Hi Kim,

That last line is the new index statistics update daemon printing its statistics on shutdown. It is cryptic and noone - myself being an exception I hope - will understand it...
I wrote this code, but since I'm on leave right now the logging code hasn't been worked on as much as it should have. I think the line is important for early stage investigation/observation in the wild, but agree that is may cause confusion.

Some improvements:
 a) Print it only if the daemon has actually done something.
 b) Document why the line is printed, i.e. for debugging/observability,  and how it can be disabled (see below). Do we already have something written down for the new feature?
 c) Document the meaning of the various metrics.

To give people a chance to judge the usefulness of this line, I'll briefly explain the metrics printed:
 - active : whether the worker thread was running when the daemon was shut down (in code: runningThread != null)
 - work : ms spent doing work
 - age :  ms since daemon creation
 - q : work units currently queued
 - p : work units processed
 - s : work units scheduled
 - k : number of known/excpected errors encountered (i.e table is dropped while being scanned)
 - u : number of unexpected errors encountered (these will affect the current work unit, but the daemon will remain active)
 - c : number of consecutive errors encountered (daemon will be disabled if a threshold is exceeded)
 - f : rejections due to a full queue
 - d : rejections due to work unit duplicates
 - o : rejections due to other causes, which is currently only attempts to schedule a work unit after the daemon has been disabled

Currently logging for the istat feature will be enabled by default, unless it has been specifically disabled by setting derby.storage.indexStats.log=false

> Update the docs to include the additional information available in derby.log. This information will be useful from product supportability stand point
> -----------------------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: DERBY-5051
>                 URL: https://issues.apache.org/jira/browse/DERBY-5051
>             Project: Derby
>          Issue Type: Improvement
>          Components: Documentation
>    Affects Versions: 10.8.0.0
>            Reporter: Mamta A. Satoor
>            Assignee: Kim Haase
>
> With DERBY-4853, we log additional information in derby.log to make it easier for the support group to work with customer problems.
> Following 2 additional properties will now always be put in derby.log (every time a database is booted) 
> user.dir 
> derby.system.home 
> user.dir is the system property which gives the user working directory. 
> If the user has not specified derby,system.home, then Derby will look for optional derby.properties file in user.dir . If user has provided derby.system.home, then the optional derby.properties will be looked in derby.system.home rather than user.dir 
> If user has not provided derby.system.home, then a sample of the 2 properties on derby.log looks as follows on my machine 
> user.dir=C:\p4clients\svnmain\client2\trunk\systest\out142 
> derby.system.home=null 
> If user has set derby.system.home to c:./temp, then a sample of the 2 properties on derby.log looks as follows on my machine 
> user.dir=C:\p4clients\svnmain\client2\trunk\systest\out142 
> derby.system.home=c:/temp 
> In addition to the above 2 properties, there can be following 3 optional properties in the log (following the 2 properties mentioned above) 
> derby.stream.error.file 
> derby.stream.error.method 
> derby.stream.error.field 
> The above 3 properties provide a way to alter where error messages get logged rather than tranditional derby.log file. In the absence of the above 3 optional properties, the location of derby.log follows the same logic as derby.properties. But if the user has set any or all of the 3 optional properties, then derby.log will not be in same location as derby.properties. In the Derby code, derby.stream.error.file takes precedence over the other 2 properties and derby.stream.error.method take precedence over the last property. 

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira