Using Query Layers with ArcGIS Server 10.0 for Java on Linux

Question

I have an ArcGIS Server 10.0 for Java running under Red Hat Linux. I would like to be able to publish a Map Document that I have created which has a working Query Layer connecting to Oracle 11g. In ArcMap the layer works as expected. I have installed all the Oracle client software on the ArcGIS Server (the server is actually running Oracle also). I am able to connect via SQL Plus as the ArcGIS server user. I have set all the environment variables to the best of my knowledge. However when I publish the Map Document I get the following in the log

WARNING The Layer:'XXXXXXXX' in Map:'Layers' is invalid. No SDEHOME environment variable set WARNING The Layer:'XXXXXXXX' in Map:'Layers' is invalid. No SDEHOME environment variable set

INFO3 Server Object instance on machine XXXXXXXX is created with errors. The Layer:'XXXXXXX' in Map:'Layers' is invalid. No SDEHOME environment variable set INFO3 Server Object instance on machine XXXXXXXX is created with errors. The Layer:'XXXXXXX' in Map:'Layers' is invalid. No SDEHOME environment variable set

I don't have ArcSDE installed and I thought the idea of Query Layers was that you didn't need it? Very confused and Google didn't turn up much :-(

Any help would be most appreciated.

Regards

Colin

• awesomo: Just checking, but I assume you have done "Enable Enterprise Geodatabase" in your Oracle DB? ArcToolbox...
• mxcolin: I'm not using SDE at all and I'm assuming I don't need to because I'm using a Query Layer? The layer works perfectly in ArcMap but not in ArcGIS Server. I'm pretty new at this so not sure how this works exactly. From what I've read using Query Layers allows you to get round using ArcSDE so I'm not sure why the error message comes up. I'm going to try it on ArcGIS Server for Windows as it wouldn't surprise me if this was a Linux issue.
• awesomo: Basically, there's two parts to SDE -- 1. All the system geodatabase tables that enable editing, topology, replication, archiving, networks, or versioning, etc. and then 2. A separate service on the database server for connections (ie port 5151 stuff, you don't use the normal database connection). Esri now says skip part 2 and just use direct connections unless there is a reason to create the SDE 5151 service.
• awesomo: Here's an article on functionality available with and without geodatabase enabled resources.arcgis.com/en/help/main/10.1/index.html#/…
• mxcolin: What a shock, works in Windows first time :-(

Answer 1

All that said above, you should be able to publish a service with data from Oracle -- which I see you just did in Windows. Anyway, you may have already done this -- but try reviewing this section in the help Making_your_data_accessible_to_ArcGIS_Server. Obviously, there is a connection problem of your linux server finding the data.

• mxcolin: Also, note I'm on 10.0 not 10.1 which might be the issue?
• mxcolin: I'm following all of what is being said here but no luck. I wonder if it is related to the 32 bit vs 64 bit versions of the Oracle client software.
• mxcolin: After much searching I found this support.esri.com/en/knowledgebase/techarticles/detail/35319
• awesomo: Did this fix the problem?
• mxcolin: So this has helped. By installing the 32 bit Oracle Client drivers the errors go away. The last message in the log is INFO3 Beginning of layer draw: XXXXXXXX however it seems to be taking an awful long time to draw the layer. I will mess around with the layer and see if I can fix it. This article was the golden nugget however.