Summary for Patch 791 (Verified)

Filtrator crashes when processing huge amount of data due to out-of-memory exception. Filtrator
should now terminate properly leaving a debug log. Also catches other unexpected exceptions.

Fix problem in Filtrator causing the last 5 minutes of every hour to be missing in the report. (No
support for hppa2 yet).

Fix in filtrator : Improve algorithm that chooses which of several simultaneous polls to use.

No support for itanium and tru64 yet.

There are two agents where one of them has restarted, and the compensations for "end-by-short-poll"
caused licenses to be "checked in" when the other agent still showed them in use. This fix will
counter that and eliminate "end-by-short-poll" compensation if the next poll for the hour is within
the poll interval time.

The configuration "archiver.compress" did not update the files in the archive automatically. This is
fixed for recorder files, but is not fixed for other files.

No support for itanium and tru64 yet.

The patch is tested on i686-linux, x86_64-linux and powerpc-aix5.1.

Fixes issues with handling of TimeZone/DST information when processing data. All
back-end processing has been updated to reduce potential problems.

Allows turning off check for error in recorderperiod archive. It will use far less time and cpu if
it is turned off. By default error-checking is on.

In /var/opt/openit/etc/daft.conf it is set by:

archiver.skip-recorder-check(Bool)="true|false"{}

Setting it to "true" turns off the error checking.

If checking remains turned on ("false" or default), it will be about twice as fast as the archiver
in patch 752. It only applies to recorderperiod data.

Fixes problems with handling of expires and valid_from parameters in
featureset map files

Update for filtrator and archvier and flab,
which had problems with speed performance.

Fix: Now sorting POLLSTART before POLLEND when in the same second.

Fix: Now ignoring status records when in poll with start time at the same time as end time of
another poll (same as if start time was before end time of the other poll).

Fix: Max in use higher than max available.
Fix: Max in use is greater than the actual license count (two licpoll client)

Fix: flexlog2 data (configured using auto configure in Windows) does not show up in reporting
server: "log file use" data type(s)

Fix: Bug introduced in patch 713, causing some data to be missing in database.

This patch include the following fixes

1. Filtrator over-reported elapsed time for winapp data

One hour got double usage everr 24 hours.

2. Fix handling of max 1 license per user/host handing of FlexNet debug data

Filtrator didn't handle the configuration in the 'flc_exception' config fil. Now it's possible
to force filtrator to never count more than a single license usage for feature/user/host
combination.

3. Force FlexNet debug log parsing to reset internal values at re-read 

Processing (flab) will now set all licenses to 0 when finding a re-read line in the debug log.
This is to match behaviour seen in debug log when licenses are gradually re-rechecked out

There is also a new option to configure a quarantine period for flexpoller data after
re-read/re-start

4. Ability to force ignore of feature version when processing data added

New parameter to force processing to ignore feature version when processing. 

Several other minor corrections dealing with conflicting input between poll data and debug
file have been added.

Fix problem in Duplicate filter where origsource internal value was not
correctly set when using the "limited" function. This only caused problems
for the 

  olap-license-total-usage (data type 102)

where "Max available licenses" could not be correctly calculated for "all"
when multiple feature versions where available.

This filtrator has also been updated to not using semaphores for locking
the debug log file.