|
Home > Archive > Apache JDO Project > August 2005 > TCK logging proposal
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
TCK logging proposal
|
|
| Michael Watzek 2005-08-16, 5:45 pm |
| Hi,
during last t-conference we discussed to improve logging of TCK runs.
Below, you find a summary of our discussion. Please comment!
Current state:
Separate log files are written by JPOX, JPOX enhancer, Derby, and TCK.
The locations of these log files are:
- JPOX: "target/jpox.log".
- JPOX enhancer: "target/enhanced/<identity type>/jpox.log"
- Derby: "target/database/derby/derby.log
- TCK: "target/logs/<timestamp>-<database>-<identity
type>-<configuration>.txt"
The current logging configuration determines that JPOX and JPOX enhancer
append output to log files. Derby truncates the log file for each TCK
configuration. In contrast, the TCK writes a new log file for each TCK
configuration result.
Problem:
Each component writes to a different log directory. The directories are
not under a common root. This fact complicates the log analysis. On the
other hand, the TCK writes a separate log file for each configuration
result to directory "target/logs". Currently, there are 14 different
configurations. There will be approximately 22 when the inheritance
proposal is implemented. Thus, each TCK run will produce 22 new files in
directory "target/logs". Obviously, this fact reduces facility of
inspection.
Proposal:
The proposal is to have separate log directories for each TCK run. These
log directories are located under the common parent directory
"target/logs". The naming pattern for these directories are given by
timestamp format "yyyyMMdd-HHmmss". The directory contents are:
- Separate log file for each TCK configuration result. The naming
pattern is: "<database>-<identity type>-<configuration>.txt".
- Separate JPOX log file for each TCK configuration. The naming pattern
is: "<database>-<identity type>-<configuration>-jpox.txt".
Thus, each TCK run will produce 44 files in separate log directories.
The proposal is to write the JPOX enhancer log to directory
"target/logs/enhancer". There, separate files will be written for each
identity type. Successive enhancer runs append the log to these files.
The file name pattern is "<identity type>-jpox.txt".
The proposal is to write the database log to directory
"target/logs/database". There, a common file for all identity types and
configurations will be written. Each run will append the log to that
file. The file name pattern is "<database>.txt"
The log level of all components is "INFO".
Regards,
Michael
--
-------------------------------------------------------------------
Michael Watzek Tech@Spree Engineering GmbH
mailto:mwa.tech@spree.de Buelowstr. 66
Tel.: ++49/30/235 520 36 10783 Berlin - Germany
Fax.: ++49/30/217 520 12 http://www.spree.de/
-------------------------------------------------------------------
| |
| Michael Bouschen 2005-08-18, 5:45 pm |
| Hi Michael,
I like the proposal. I added a few comments plus a suggestion for an
extension below.
> Hi,
>
> during last t-conference we discussed to improve logging of TCK runs.
> Below, you find a summary of our discussion. Please comment!
>
> Current state:
> Separate log files are written by JPOX, JPOX enhancer, Derby, and TCK.
> The locations of these log files are:
>
> - JPOX: "target/jpox.log".
> - JPOX enhancer: "target/enhanced/<identity type>/jpox.log"
> - Derby: "target/database/derby/derby.log
> - TCK: "target/logs/<timestamp>-<database>-<identity
> type>-<configuration>.txt"
>
> The current logging configuration determines that JPOX and JPOX enhancer
> append output to log files. Derby truncates the log file for each TCK
> configuration. In contrast, the TCK writes a new log file for each TCK
> configuration result.
>
> Problem:
> Each component writes to a different log directory. The directories are
> not under a common root. This fact complicates the log analysis. On the
> other hand, the TCK writes a separate log file for each configuration
> result to directory "target/logs". Currently, there are 14 different
> configurations. There will be approximately 22 when the inheritance
> proposal is implemented. Thus, each TCK run will produce 22 new files in
> directory "target/logs". Obviously, this fact reduces facility of
> inspection.
I agree, this does not work anymore, too many log files in that directory.
>
> Proposal:
> The proposal is to have separate log directories for each TCK run. These
> log directories are located under the common parent directory
> "target/logs". The naming pattern for these directories are given by
> timestamp format "yyyyMMdd-HHmmss". The directory contents are:
>
> - Separate log file for each TCK configuration result. The naming
> pattern is: "<database>-<identity type>-<configuration>.txt".
> - Separate JPOX log file for each TCK configuration. The naming pattern
> is: "<database>-<identity type>-<configuration>-jpox.txt".
Sounds good.
>
> Thus, each TCK run will produce 44 files in separate log directories.
Just for my understanding: a TCK run will create one directory e.g.
target/logs/20050818-101723. This directory contains all the logs files
from the tck run including the TCK run results and the JPOX log output.
The JPOX enhancer logs and the database log file do not go into this
directory, they have their own directory, correct?
How about adding a new file showing a summary of the tck result for the
different configurations. It would include two lines per configuration:
- The first line shows the database, identitytype and the name of the
configuration.
- The second line lists the test results as we have it today at the end of
the tck output:
Tests run: 396, Failures: 17, Errors: 25, Time: 281,124 seconds.
This makese it easier to compare the results of this tck run with a
prvious run. What do you think?
>
> The proposal is to write the JPOX enhancer log to directory
> "target/logs/enhancer". There, separate files will be written for each
> identity type. Successive enhancer runs append the log to these files.
> The file name pattern is "<identity type>-jpox.txt".
>
> The proposal is to write the database log to directory
> "target/logs/database". There, a common file for all identity types and
> configurations will be written. Each run will append the log to that
> file. The file name pattern is "<database>.txt"
>
> The log level of all components is "INFO".
Sounds good.
Regads Michael
>
> Regards,
> Michael
> --
> -------------------------------------------------------------------
> Michael Watzek Tech@Spree Engineering GmbH
> mailto:mwa.tech@spree.de Buelowstr. 66
> Tel.: ++49/30/235 520 36 10783 Berlin - Germany
> Fax.: ++49/30/217 520 12 http://www.spree.de/
> -------------------------------------------------------------------
>
--
Michael Bouschen Tech@Spree Engineering GmbH
mailto:mbo.tech@spree.de http://www.tech.spree.de/
Tel.:++49/30/235 520-33 Buelowstr. 66
Fax.:++49/30/2175 2012 D-10783 Berlin
| |
| Michael Watzek 2005-08-19, 7:46 am |
| Hi Michael,
> Hi Michael,
>
> I like the proposal. I added a few comments plus a suggestion for an
> extension below.
>
>
>
>
> I agree, this does not work anymore, too many log files in that directory.
>
>
>
>
> Sounds good.
>
>
>
>
> Just for my understanding: a TCK run will create one directory e.g.
> target/logs/20050818-101723. This directory contains all the logs files
> from the tck run including the TCK run results and the JPOX log output.
> The JPOX enhancer logs and the database log file do not go into this
> directory, they have their own directory, correct?
Correct. See below.
>
> How about adding a new file showing a summary of the tck result for the
> different configurations. It would include two lines per configuration:
> - The first line shows the database, identitytype and the name of the
> configuration.
> - The second line lists the test results as we have it today at the end of
> the tck output:
> Tests run: 396, Failures: 17, Errors: 25, Time: 281,124 seconds.
>
> This makese it easier to compare the results of this tck run with a
> prvious run. What do you think?
This is a good idea! What about "TCK.results" to be used as a file name.
Concerning the format of the two lines per configuration, I suggest to
have the file name of a particular configuration result as the first
line, e.g.:
derby-app-alltests.txt:
Tests run: 396, Failures: 17, Errors: 25, Time: 281,124 seconds.
I think, we need a final TCK output written to the console at the end of
a TCK run, e.g.:
"All configuration passed." or
"5 configurations failed. See file
'target/logs/20050818-165457/TCK.results' for details."
Regards,
Michael
>
>
>
>
> Sounds good.
>
> Regads Michael
>
>
>
>
>
--
-------------------------------------------------------------------
Michael Watzek Tech@Spree Engineering GmbH
mailto:mwa.tech@spree.de Buelowstr. 66
Tel.: ++49/30/235 520 36 10783 Berlin - Germany
Fax.: ++49/30/217 520 12 http://www.spree.de/
-------------------------------------------------------------------
|
|
|
|
|