yf report broadcast issue
26 February, 2015
We have a scenario in which we are trying to broadcast a report to 900+ users. On doing this, we get a DBAction.commit() Exception. (see Screenshot attached).
1. what could the resolution for this issue?
2. Is there a way to find out how many and which users the report was broadcasted to before it failed?
3. On A similar note, can we broadcast the report to a local and / or network folder? (not FTP)
Please note that if we broadcast the report to 30 users, the report broadcasts successfully. We tried increasing the JDBCTimeout in the web.xml, also tried increasing the connection pool size, but it didn�t help.
Also, is there a performance tuning guide which can be used for defining below parameters?
for example
What is the ideal...
� DB Connection Pool Size
� JDBC TimeOut
..as the number of users increase?
1. what could the resolution for this issue?
2. Is there a way to find out how many and which users the report was broadcasted to before it failed?
3. On A similar note, can we broadcast the report to a local and / or network folder? (not FTP)
Please note that if we broadcast the report to 30 users, the report broadcasts successfully. We tried increasing the JDBCTimeout in the web.xml, also tried increasing the connection pool size, but it didn�t help.
Also, is there a performance tuning guide which can be used for defining below parameters?
for example
What is the ideal...
� DB Connection Pool Size
� JDBC TimeOut
..as the number of users increase?
Hello Nitin,
Sorry for the delay in responding to your query. We are hoping to have some information
for you soon.
Thank you for your continued patience.
Kyle
Sorry for the delay in responding to your query. We are hoping to have some information
for you soon.
Thank you for your continued patience.
Kyle
Hi Nitin,
Sorry for the delays.
When a broadcast is sent to a user, it runs the report.
If you send the report to 3 users, it runs the report 3 times.
That means that if the report takes 5 secs to run , and you broadcast to 3 users. The broadcast requires a connection to the datasource to be open for 15 secs.
If you're broadcasting to 900 users, then you need to ensure that the total run time is supported by the data source timeout.
The good news is that the 'secondary connection pool' is used for system tasks such as broadcasts and caching filters.
So you just need to increase the secondary connection pool size for this datasource to a number that will allow all 900 reports to run.
Please let me know if you have any questions or concerns on this.
Of course, if we really wanted to investigate this issue, we would also want to see the log files under Yellowfinappserverlogs.
Thanks,
David
Sorry for the delays.
When a broadcast is sent to a user, it runs the report.
If you send the report to 3 users, it runs the report 3 times.
That means that if the report takes 5 secs to run , and you broadcast to 3 users. The broadcast requires a connection to the datasource to be open for 15 secs.
If you're broadcasting to 900 users, then you need to ensure that the total run time is supported by the data source timeout.
The good news is that the 'secondary connection pool' is used for system tasks such as broadcasts and caching filters.
So you just need to increase the secondary connection pool size for this datasource to a number that will allow all 900 reports to run.
Please let me know if you have any questions or concerns on this.
Of course, if we really wanted to investigate this issue, we would also want to see the log files under Yellowfinappserverlogs.
Thanks,
David
Hi David
When we run a schedule for a large number of users and if the broadcast fails, we are currently unable to figure out which users received the report and which users were still pending at the time of failure. Is there any way to obtain this information ?
- Shashi
When we run a schedule for a large number of users and if the broadcast fails, we are currently unable to figure out which users received the report and which users were still pending at the time of failure. Is there any way to obtain this information ?
- Shashi
Hello Shashi,
There is a configuration setting in Admin/Email/General settings that allows an Admin
to receive and email when a broadcast has failed, and this email will include the recipients
that did not successfully receive the Broadcast.
I had a quick test on this and confirmed that the system does send an email and it included
the recipients that did not get the Broadcast email.
This is probably worth testing for the scenario that you have outlined.
Thank you,
Kyle
There is a configuration setting in Admin/Email/General settings that allows an Admin
to receive and email when a broadcast has failed, and this email will include the recipients
that did not successfully receive the Broadcast.
I had a quick test on this and confirmed that the system does send an email and it included
the recipients that did not get the Broadcast email.
This is probably worth testing for the scenario that you have outlined.
Thank you,
Kyle
Thanks Kyle. It appears that this feature is only available in version 7.2 . We are currently running 6.3. Is there any way to achieve the same in 6.3 - maybe through log files or something ?
- Shashi
- Shashi
Hi Shashi,
This functionality was added to the 7.0 product, forward.
Without updating your instance, your only option is to review
the YF log files, specifically the email log. A failed Broadcast
should show up in this log file.
I hope this all helps.
Kyle
This functionality was added to the 7.0 product, forward.
Without updating your instance, your only option is to review
the YF log files, specifically the email log. A failed Broadcast
should show up in this log file.
I hope this all helps.
Kyle
Hi
We tried using the secondary connection pool with a min size of 100 connections and max size of 200 connections , refresh time of 1 hour and timeout as 1 day
We scheduled our reports to run in batches of 100 ; while the first few batches went through , the fifth one again failed. The jdbc log reported non availability of a free connection.
Any suggestions ?
Also, does Yellow Fin provide an API if we had to write our own report broadcasting job ; that way we can programmatically handle such exceptions
regards
Shashi
We tried using the secondary connection pool with a min size of 100 connections and max size of 200 connections , refresh time of 1 hour and timeout as 1 day
We scheduled our reports to run in batches of 100 ; while the first few batches went through , the fifth one again failed. The jdbc log reported non availability of a free connection.
Any suggestions ?
Also, does Yellow Fin provide an API if we had to write our own report broadcasting job ; that way we can programmatically handle such exceptions
regards
Shashi