rwright
5th April 2002, 18:31
We are getting...
"1,0,0 ddc_monitor: Function server timeout"
from one of the "hours accounting" function servers.
What could be causing this?
Suggested fixes?
Also, I tried to access ftp.baan.com to get ddc documentation that was referenced in a different thread and could not attach.
bmadhu
5th April 2002, 18:35
Which middleware you are using to connect to DDC ?
rwright
5th April 2002, 18:52
Connectware
bmadhu
5th April 2002, 18:55
Try increasing the front end time out before starting your DDC Server. You may have to contact Connectware for the specific settings. One more way is to set the DDC_MONITOR_OFF environment variable before starting DDC Server. This way you wont get a time out from DDC Monitor but if there is no response from the function server then DDC will hang.
reldif
6th April 2002, 01:02
Richard,
Is this for Baan IV or Baan V?
I would run the test function server (tuddc2350m000) for the hours accounting function that is in question, with values that caused the "Function Server Timeout". You will probably see a message box or something pop up. Connectware's DDC driver will not be able to handle this, but the message should tell you what is going on. If this is the case, you may have to contact Connectware (well, Epic Data) or get a function server consultant to supress the message that is poping up. If this is a standard Baan function server, you may want to check with Baan to see if they have an updated function server.
If you do not see a message and the function server is just taking way too long, then you will have to either change the amount of time it takes for the ddcmonitor to shut down your function server, or turn off the monitor completely. Madhu already told you how to turn it off with the DDC_MONITOR_OFF env. variable, but you can also mess with the "time" using MONITORDEFAULT variable. Baan V has modified this and the environment variable is DDC_MONITOR_ON instead. The default for MONITORDEFAULT is 12000 milliseconds. You should be able to modify and set these environment variables in you bnddc.sh script. But, if you try this and you get a message box, then the DDC driver will hang.
hope this helps,
mike
rwright
12th April 2002, 16:39
This is Baan IVc4. The "hours_accounting" function server is being used as an employee timeclock at this location. At the beginning and ending of a shift, there are a lot of transactions which is when the problems occur. The problems are slow response and timeouts. Apparently there is only about 50% success rate during these times. When you have 50 employees waiting in line to clock in, this is the wrong time to slow down to minute+ response.
I understand that this is not the purpose of the "hours_accounting" function server, but we need to get this working or switch to a better solution.
What are our options to get this resolved?
lbencic
12th April 2002, 20:07
Have you tried running this function through test function server at the same time your shifts are clocking out? Test function server does not use the ddc server portion of ddc, so you would be able to tell if the problem is in the ddc server being clogged or if it is a problem with the performance / program not responding quickly enough.
My company does ddc development. We have many clients using our version of hours accounting. We have found that for certain heavy use/shift oriented clients, performance is a problem. We continue our ongoing efforts to speed this program up. The standard one will not cut it.
Here's what we have done:
Break the functions into separate function servers (currently only the Baan V version)
Streamlined code to not repeat validations (standard was full of this)
Modified ddc code to handle errors / timeouts more effectively.
Work with companies to determine the best time out / retry options for their system. You need to give it enough time to finish. If it's not finishing, you will get time outs all over the place, and retries will only clog it up more.
For 1 client, this was still not enough, and we had to modify our middleware a bit to support off-line processing of the clock in / clock out transactions. This allows the user to enter their transaction and the terminal is immediatly available for the next user. Any errors are stored in the standard ddc off-line error log. There should be few errors when an employee clocks out (open jobs in our version are closed on clock out), so there is very little problem with the user not seeing any messages returned. The IS department checks the logs once a day and communicates to the shift supervisors if there has been a problem.
This is not a sales pitch, I would think that Connectware has sold this function server themselves many time, and should have some suggestions for you. The off-line processing especially has solved the problem finally for our heavy use client. Have you checked with them? If they are not helping, you can always check with us :) Our ddc code works with any middleware (the split functions, and the streamlined validations), but the off-line processing and ddc code modifications are part of our middleware.
Good Luck
Lisa Bencic
RMCis - www.rmcis.com
lbencic@rmcis.com