In September we put a new onDemand 8.4 server in to replace or 7.1.2.6 server. We began prodution on the new server at 8.4.0.3 and immediately notice that we would up to 100+ plus load failures a night. We have opened an PMR with Support but they cannot tell us why the Loader is not removing the .ARD file after a successful load. IBM suggested that we apply 8.4.1.0 fix pack and it did fix some issues but not this one. These particular loads that are failing are coming from a Windows Image capture system (TIFF Files)and load at 6:00PM ever night in the 100's. Ofcousre each load contains a .IND, .OUT, and ARD files. What were are seeing is that the load will be successful, the .IND and .OUT file will be remove from the ARSLOAD directory but .ARD trigger file will remain causing the a second load to fail 1 second later. There could be 20 to 200 of these failures a night all a second after loading successfully. It is like the System is out pacing it self and seeing the previous trigger file before it has bee removed but then failing with the error that this file has not been indexed. Has anyone seen or heard of this problem. Also I cannot duplicate this issue from command line.
Pre-Thanks,
waytec
A second load attempt 1 second later sounds quite quick. Is the first load even completed that soon?
Is the .ARD file getting renamed to .failed ?
Some are a second later and some are 1 minute later. And yes the remaining .ARD is renamed to .ARD.Failed What do you think? ???
Hi there.
Which platform are you on?
Also, what is the 'polling time' you have set on arsload (the -t option)? You don't have the -n option set, do you?
Can you write a little script to do the loads for you? On a UNIX platform, you could use this little script in the interim:
#!/usr/bin/ksh
for i in `ls *.ARD`
do
arsload <parameters> $i
done
(This script assumes that arsload invoked with a file name *is* removing the file afterwards. If it isn't, add a 'gzip $i' on a separate line after the arsload line, and before the 'done' line to compress files. You can clean out the directory with a simple 'rm *.gz' when your disk starts getting full.
Hope this helps!
-JD.
What Up J.D., We are running OD MP 8.4.1.0 on Windows 2003 SP3. The Arsload interval is set to default 600 seconds.
Erk.
Okay, it's fairly obvious that my UNIX shell script isn't going to help you out there...
I'd definitely call this one a bug -- can you show us some error messages that arsload is kicking out?
-JD.