I work at a very large organisation where we by default use proxy servers for all our users. Basically every browser request is routed through a proxy server unless there is an exception in the “proxy.pac” file for it. The proxy.pac files is picked up by reading it automatically from the proxy server itself, i.e. we don’t maintain it locally on each workstation.
We use Infoview (Apache 2.0 / Tomcat 4.1 on Win2000) for a wide range of reporting requirements. This of course includes some very long-running queries, e.g. some reports can as long as half an hour or more to run. For such reports, users are hitting timeouts and getting “ERROR: The requested URL could not be retrieved”. We’ve hauled out the usual suspects, changing the Apache timeout, the “timeout for interactive batch”, etc. settings, and we’ve checked the Microsoft support article about browser timeouts (we use IE6 so the timeout is supposedly 60 minutes).
We’re fairly certain that the problem lies with the proxy server. When a user runs a large report, Infoview is off in never-never land waiting for the WebI to send results back, and WebI is probably stuck waiting for the database to return something. After around 5 minutes, the proxy server returns an error to the client browser even though things are chugging along fine as far as Infoview / WebI are concerned.
Does this sound possible, or does Infoview already implement some sort of keep-alive such that timeouts like this should not happen? If this is possible, can anyone think of a way to work around it? Our Network Services team are reluctant to add more entries to the proxy.pac file for our Infoview websites as the size of this file is already too big and it’s causing performance problems. Basically I’m hoping to get Infoview to work with long-running reports, even when requests are going through a proxy server with a relatively short time-out.
Regards,
David E. Hansen
BO E6.5.1 / Win2000 / Apache 2 / Tomcat 4
David, this is probably a looong shot but ===> if you get into WebI Admin and go to WISessionManager, what is the setting for Session timeout?
We were having a similar issue with users not being able to complete long running queries… except in our case they were timing out at 60 minutes rather than the 30 you’re seeing. I bumped it up to 90 minutes (5400 seconds) and life is good once again – or at least they’re yelling at me for different stuff now.
In our case, if a query runs longer than 90 minutes, I require them to bust it up into smaller, more manageable queries OR I research the query’s sql to see what tuning/mods I could do on the dB.
I don’t think InfoView has any kind of keep-alive mechanism. We ran into a similar issue when we tried to implement load balancing using a Cisco router. Our Network group insisted that all requests run through proxy servers between the router and the WebI servers. The proxy servers had a 5-minute timeout setting that the Network group refused to change.
Because of the timeout, all long-running queries were failing in WebI and ZABO. Our workaround was to help users login directly to one of the servers, instead of going through the router. Of course, this defeated the purposes of load balancing. We opened a case with BO, but couldn’t find a way for BO to keep the session alive on the proxy server. Eventually we pulled the router out and won’t do any load balancing until we install some new hardware next year.
For the very long reports, do not use IE.
Use vbscript to call the reports (and fill prompts if required), with vbscript you can determine the timeout setting which i think you cannot do with a browser.
We have a similar issue at our place. This has nothing to do with any of the setting on webi or infoview. We have a session time out of 3 hours. The user gets the proxy error in say 5 minutes. This has to do with the proxy time out settings.
Luckily the users having this issue are understanding enough to uncheck the proxy script when using Infoview and then recheck it back for other things. I know its a pain but they are willing to go to that step for using BO.
Lately, our ZABO user gets the Exception CS, Job Already in use error when they try to run a large query. I opened a ticket twice with BO, they said its the proxy time out problem. We have increased the setting for the following: Clients 180 sec or 3 minutes,
Servers 180 sec or 3 minutes, and Refresh 90 sec or 1.5 minutes. But the user still gets the same error. Any suggestions on this will be greatly appreciated. We use IE and UNIX for WEBI server.
Sharon,
Are you sure that users using ZABO have a proxy time out issue?
I was told by BO support that the time out issue doens’t apply for ZABO users. Huh!! BO Support is getting to a point “This is RIDICULOUS”
Yes, It’s for ZABO user. I tested in both ZABO and Full Client, the full client ran fine for the same report. I also tried to see if theWI session Manger parameter has anything to do with it. But it runs fine in DEV with the WI sessionManager setting to 3000 sec only. I were the only user who logged in the ZABO server. Out prod has 7000 sec. So I don’t think it’s anything to do with the WI SessionManager. BO still didn’t have an answer for this. It’s a big issue to us since we have more ZABO and WEBI users than full client. What else I can try? Normally, we have 15 concurrent users log in prod. Thanks!
Sharon,
When we had the issue it was due to a proxy setting timing limit which happens to be on a browser. So I would go to internet oprions > connections > Lan Settings and uncheck the proxy.
But since you are saying that in your case its ZABO users and not Infoview users I am not sure what the resolution would be. Did you try talking to your network guys about this proxy setting limit!! They might help you.