Due to a problem specific to Windows server, all components in the iProcess Engine can hang when trying to maintain a consistent synchronized accurate time between them.
Article ID: KB0092584
Updated On:
| Products | Versions |
|---|---|
| TIBCO iProcess Engine (SQL) | - |
| Not Applicable | - |
Description
Description:
Note: This Known Issue applies to the TIBCO iProcess Engine 10.3.0.09 or later and 10.5.0 or later up to 10.5.0.13 versions running on Windows and being used with Oracle 9i, Oracle 10g or SQL Server databases.
================================================
All processes on the TIBCO iProcess Engine on Windows can go into a sleep state, for an indefinite period of time.
No DEBUG can be applied and commands such as ‘plist –p’ will not return.
If you see this symptom it can be confirmed by reviewing the $SWDIR/tsys/basetime file for inconsistencies, this must be done on a system that is exhibiting the problem, it cannot be carried out after a restart.
The basetime file has the following format:-
a:b:c:d:e:f:g:h:i
where a to i are decimal numbers.
Calculate the difference between the values of the 1st [a] and 6th [f] fields in the basetime file.
Then calculate the difference between the values of the 3rd [c] and 8th [h] fields in the basetime file. Now compare these too differences they should be broadly similar, particularly when compared to the scale of the original numbers (which will tend to be very large numbers), if the 2 values are different by a large amount i.e. the second figure is a ratio of 10 times greater or more than the first number this would indicate that this problem has occurred.
Users running version 10.5.0 on Windows should apply the hotfix 10.5.0.13.1.
Those running on 10.3.0 should apply hotfix 10.5.0.13.1 (which is the same hotfix as needed for 10.5). Version 10.6.0 is not affected by this issue.
Workaround:
There is no guaranteed workaround for this issue and applying the hotfix is recommended.
See Related LBNs for hotfix information.
However the likely chance of this problem occurring can be reduced.
This can be achieved by reducing the frequency with which the sentinels check for the time drift, which is the operation that can cause the problem.
This can be done using the WINTIME_RESYNC_PERIOD process attribute. This is set to 300 by default, to get it to check every 300 seconds for time drifting.
Setting it to check every hour, will reduce the chance of the problem happening. We recommend that the value is set to 3600 to still try to avoid too much drift, but to lower the chance of the problem occurring.
Customers experiencing this problem should request that they are supplied a version of the server or the hotfix with CR30137 in it.
Symptoms:
Cause:
Note: This Known Issue applies to the TIBCO iProcess Engine 10.3.0.09 or later and 10.5.0 or later up to 10.5.0.13 versions running on Windows and being used with Oracle 9i, Oracle 10g or SQL Server databases.
================================================
All processes on the TIBCO iProcess Engine on Windows can go into a sleep state, for an indefinite period of time.
No DEBUG can be applied and commands such as ‘plist –p’ will not return.
If you see this symptom it can be confirmed by reviewing the $SWDIR/tsys/basetime file for inconsistencies, this must be done on a system that is exhibiting the problem, it cannot be carried out after a restart.
The basetime file has the following format:-
a:b:c:d:e:f:g:h:i
where a to i are decimal numbers.
Calculate the difference between the values of the 1st [a] and 6th [f] fields in the basetime file.
Then calculate the difference between the values of the 3rd [c] and 8th [h] fields in the basetime file. Now compare these too differences they should be broadly similar, particularly when compared to the scale of the original numbers (which will tend to be very large numbers), if the 2 values are different by a large amount i.e. the second figure is a ratio of 10 times greater or more than the first number this would indicate that this problem has occurred.
Users running version 10.5.0 on Windows should apply the hotfix 10.5.0.13.1.
Those running on 10.3.0 should apply hotfix 10.5.0.13.1 (which is the same hotfix as needed for 10.5). Version 10.6.0 is not affected by this issue.
Workaround:
There is no guaranteed workaround for this issue and applying the hotfix is recommended.
See Related LBNs for hotfix information.
However the likely chance of this problem occurring can be reduced.
This can be achieved by reducing the frequency with which the sentinels check for the time drift, which is the operation that can cause the problem.
This can be done using the WINTIME_RESYNC_PERIOD process attribute. This is set to 300 by default, to get it to check every 300 seconds for time drifting.
Setting it to check every hour, will reduce the chance of the problem happening. We recommend that the value is set to 3600 to still try to avoid too much drift, but to lower the chance of the problem occurring.
Customers experiencing this problem should request that they are supplied a version of the server or the hotfix with CR30137 in it.
Symptoms:
Cause:
Issue/Introduction
Due to a problem specific to Windows server, all components in the iProcess Engine can hang when trying to maintain a consistent synchronized accurate time between them.
Feedback
Yes
No
Powered by
