on 03-01-2010 8:12 AM
hi all
i am attempting to run a large scheduling run of 3550 report instances. this is obviously going to take some time so is difficult to test within a decent time frame (3.5 days total - yes they are big)
my problem is that some (20%) of the reports remain in a PENDING state.
the code used to configure the schedule timing is
setRightNow(true)
setType(CeScheduleType.ONCE)
the instance detail for one of the pending jobs is
creation time: 2/26/2010 6:05 pm
next run time: 2/26/2010 6:05 pm
both the invoking java application and the enterprise server are on the same time, locale, etc
can anyone help me out?
cheers
warwick
Edited by: warwick.mayson on Mar 1, 2010 9:13 AM bad grammer
I'm assuming you're not looking for simultaneous runs, so why not stagger the schedule start times, rather than use Right Now?
These are Crystal Reports, yes?
Sincerely,
Ted Ueda
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
hi ted
thanks for your answer. yes, crystal reports.
if i stagger the start time of each job will it not potentially make the run take longer? my customer is asking for a dramatic increase in speed and i hope not to exacerbate the problem. though, its is better to get results than not
could you explain why the scheduled jobs get hangs in the pending state and not recover? perhaps i could develop a strategy based on this information
will setting IScheduleInfo.setBeginDate() of each report to a different value help?
thanks again
warwick
You aren't going to get 3550 jobs running simultaneous, anyways. Job Servers work by spawning child processes for report generation, and the maximum number of child processes are configurable in the CMC.
There's two issues that's commonly encountered, of which one may apply to you:
1. If the Enterprise Java SDK application is hosted on a different machine from the CMS service, you have to ensure that the two machines have their times synched using a time server. When you schedule a report to run Right Now, the Right Now time is determined from the SDK app machine. So if the times are out of synch, a job may wait in pending.
2. Scheduling Web Intelligence jobs, you may encounter a problem where the Web Intelligence Job Server may respond very slowly when so many scheduled jobs are requested at once.
If you set the schedule to not run "Right Now" but Once, where you stagger the start time of jobs, and the issue still occurs, I recommend opening a support case with SAP.
Sincerely,
Ted Ueda
User | Count |
---|---|
69 | |
26 | |
9 | |
9 | |
7 | |
6 | |
4 | |
4 | |
4 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.