Common problems and solutions |
|
Release 9.3 |
Below are listed some questions or issues that you may face when working with ArcGIS Serverand some recommended solutions. If you don't find the problem you are looking for, you can also search for articles on the ESRI Support Center.
The .NET and Java versions of ArcGIS Server can coexist on the same machine. If you install them together, you will need to follow the procedure below to ensure that your server directories correctly map to the virtual directories appropriate for your IIS or Apache Web server:
The server object manager (SOM) and server object container (SOC) accounts are used internally by the GIS server and need only have limited permissions on the machine. Chances are you'll only encounter them when you're installing ArcGIS Server on other machines, or when you're giving the GIS server permissions to access your data. In most cases, it's sufficient to use the default account names suggested by the postinstall wizard (ArcGISSOM and ArcGISSOC) and let the wizard create the accounts for you. The postinstall wizard creates local accounts, which are recommended over domain accounts for security reasons.
ArcGIS Server names cannot exceed 15 characters or you will encounter errors during the postinstall and when trying to connect to the server. This is a known limit related to a Microsoft specification for Active Directory names.
The installation is not automatically logged, although if you run the installation from the command line you can include parameters that will create a log file. See the ArcGIS Server Install Guide for help with installing from the command line.
You can view which features of ArcGIS Server are installed through the Add or Remove Programs dialog box in Windows. Click the Change button under ArcGIS Server for the Microsoft .NET Framework, then choose Modify to view or change the list of installed components.
When you uninstall and reinstall ArcGIS Server, your server and service configuration files are preserved. However, the web.config and Tasks.xml files are removed by the uninstallation process. If you manually edited these files, you will need to make a backup copy before uninstalling, then copy the information back in after reinstalling.
Preserving security information in the web.config fileIf you added security information for your Web services, these were written to the web.config file and need to be preserved. Copy your web.config file to a location unrelated to ArcGIS before you uninstall ArcGIS Server. You can find the web.config file in <Virtual root directory>\<ArcGIS Server instance name>\Services (Example: C:\Inetpub\wwwroot\ArcGIS\Services\web.config). After re-installing, copy and paste just the security-related tags back into the new web.config file.
Preserving information about custom tasksTasks.xml is used for integrating tasks that you've developed into the Manager user interface, so you only need to back up this file if you've developed custom tasks. You can find Tasks.xml in <Virtual root directory>\<ArcGIS Server instance name>\Manager\Applications\App_Data (Example: C:\Inetpub\wwwroot\ArcGIS\Manager\Applications\App_Data\Tasks.xml). Before uninstalling, copy this file to a location unrelated to ArcGIS. After reinstalling, replace the new Tasks.xml file with the old Tasks.xml file.
If you receive the message "Your ArcGISServer license has expired", please contact ESRI Customer Support to obtain a new license, then re-run the "Authorize ArcGIS Server" portion of the GIS Server Post Install.
If you have set up a distributed system where the Web server, server object manager (SOM) and server object containers (SOCs) are all on separate machines, ArcGIS Server will not function without the network. If, however, all these components are running on the same physical machine, ArcGIS Server will operate correctly as long as all data is referenced using local pathnames rather than shared network directories with UNC pathnames. For example, when you publish a resource such as a map document, publish it from your C:\ drive and ensure that all layers reference local data through local pathnames as well.
Additionally, if the computer is part of a domain and you have disconnected the computer from the domain, you must use local accounts for the SOM account, SOC account, ArcGIS Web Services account, and the account you use for logging into Manager.
When using ArcGIS Server in a Windows Workgroup environment (not to be confused with ArcGIS Server Workgroup), there are a few important steps you need to follow to make sure that GIS server authentication will function correctly. See the section "Network environments" in the topic How the GIS server works for instructions.
Occasionally, the situation may arise that your data is on a machine with no components of ArcGIS Server installed and you are following the recommended practice of using local accounts for the SOC account. You will need to use the operating system tools to create a local SOC account on the machine containing your data. Create a local account on the machine hosting your data and assign it the same name and password as the SOC account on all the other machines in your deployment. The GIS server will then be able to recognize that it has permissions to access your data.
Failover and round-robin are techniques to provide a backup server in case one server in your configuration goes down. As you design your Web applications in Visual Studio, you can specify additional GIS servers that the application will use and whether they will act in a failover or round-robin mode. Failover and round-robin can also be implemented programmatically using the Connection Library included with the ADF.
You are asked to enter the ArcGIS instance name when you install ArcGIS Server. If someone else performed the installation or you cannot remember what you entered, there is an easy way to tell what the instance name is:
From the Windows Start menu, navigate to Start > All Programs > ArcGIS > ArcGIS Server for the Microsoft .NET Framework. If you see the shortcut ArcGIS Server Manager then you have the default instance name, which is ArcGIS. If you have some other instance name, it will be appended to the Manager shortcut in parentheses, for example: ArcGIS Server Manager (Instance1).
Each instance has its own folder in the C:\Inetpub\wwwroot directory. If you can navigate to C:\Inetpub\wwwroot\ArcGIS, then your instance name is probably the default of ArcGIS.
ArcGIS Server uses two ArcSOC.exe processes internally for logging and directory management. You will still be able to see these processes running in Windows Task Manager even if you stop all your services in Manager or ArcCatalog.
Additionally, geoprocessing services use multiple ArcSOC.exe processes to manage scheduling and status of jobs. For each geoprocessing service configuration, you will see one ArcSOC.exe process, plus two ArcSOC.exe processes for each running instance. For example, if you have one geoprocessing service with three running instances, you will see seven ArcSOC.exe processes. These processes will go away when you stop the service or shortly thereafter.
When using ArcCatalog to administer your ArcGIS Server, you may get an error message if you attempt to use the "?" button to get help. This happens on the Windows Vista operating system. To see the help, you'll need to download the Windows Help program (WinHlp32.exe) for Windows Vista.
To log in to Manager, you must use an account that is a member of the agsadmin group on the server object manager (SOM) machine. If you want to create Web applications with Manager, the account must also be an Administrator on the Web server machine.
Logging in to Manager in Windows XP also requires that simple file sharing be disabled. In the simple file sharing model, all attempts to log on to the computer from across the network are forced to use the Guest account. Manager and Web ADF applications need to authenticate as the ArcGIS Web Services account and not the Guest account. Therefore, you will need to disable simple file sharing if it is not disabled already. To disable simple file sharing, follow these steps:
If the web server name hosting the Manager application contains underscore characters in the name, attempting to log in to Manager in an Internet Explorer browser will fail without any error message.
Options for working around this issue are:
If you have performed a distributed installation of ArcGIS Server where the components of ArcGIS Server are running on more than one machine, you will need to change the default log file location from a local path to a UNC path pointing to a shared directory that all the machines in your ArcGIS Server system can access. Although Manager contains an interface for changing the log file location, it is a known issue at 9.2 that you must use ArcCatalog to initially change the log file location after you perform a distributed installation. See Specifying the log file location (ArcCatalog) for instructions.
You will also need to grant the SOM and SOC accounts Read and Write permissions to the new log directory.
At the ArcGIS Server 9.3 final release, it is a known problem that when you restart the ArcGIS Server Object Manager (SOM) service, the server is depicted as being offline the next time you log in to Manager. As a result, many options in Manager are unavailable (grayed out).
You can make the options available again by closing Manager and re-opening it.
Using ArcGIS Server 9.3, you can publish Web Map Services (WMS), Web Feature Services (WFS), and Web Coverage Services (WCS).
Learn more about OGC support in ArcGIS Server.
To understand how to configure the correct permissions required for ArcSDE and other geodatabase layers, see Preparing resources for publishing.
If you see a blank Preview tab with coordinate values appearing below as you move the mouse, it's likely that ArcCatalog can't get the map image from the virtual directory you have associated with your output directory. In this situation, ArcCatalog has all the information about the map except for the actual image, which is why you see the coordinates as you move the mouse. The best chance of fixing the problem is to check the virtual directory settings using your Web server administration software, making sure that the virtual directory is correctly pointing to the output directory on disk.
If you want to verify that the problem is with the output directory, adjust the map service's properties so that the supported image return type is MIME only. This setting does not use an output directory. If you see the image with MIME only and you don't see it with MIME + URL, then you know there is a problem with the output directory and/or the virtual directory.
You can browse to the output directory on disk to make sure that the images are being created inside. If you see images being added to the output directory as you try to Preview the map service, then the problem is with the virtual directory settings.
The amount of time needed to create your map cache depends on the type of cache you are building (fused or multilayer), the scale levels you have chosen, and the amount of server resources you have dedicated to generating the cache. These factors are discussed in detail in the section Anticipating cache creation time in the topic "Planning a map cache."
You can copy your <ArcGIS>\server\user\cfg folder from the development machine and paste it on the production machine in the same location, overwriting the original cfg folder. You will then need to manually restart the ArcGIS Server Object Manager Windows service. If your services' data resides in a different folder structure on the production machine, you will need to modify the service properties to reference the new paths to the data.
Using nonpooled services over an ArcGIS Server Internet connection is not recommended; instead, use an ArcGIS Server local connection.
Internet connections use the services in a purely stateless fashion. Each request to a service via the Internet results in a context acquisition and release on the back-end DCOM component of the service. If the service is nonpooled, each context release results in the destruction and re-creation of a service instance. This can be computationally expensive, especially if a client is making multiple requests to the service. For example, a simple Web ADF mapping application might connect to the service to get the Map object, the MapDescription, and the ImageDescription, then execute a draw request. Each of these operations involves a context acquisition and release. If the service is nonpooled and the client is connecting via the Internet, each of these operations results in the creation and destruction of a service instance, thereby slowing performance.
If you update an area of a map or globe cache, users of ArcMap, ArcGlobe, and ArcGIS Explorer who have already visited that area and extent must clear their local image caches before they can see the updates. As a server administrator, you need to alert your users when updated data is available so that they know to clear their caches.
See Using a map cache to learn how to clear the cache in these applications.
By default, ArcGIS Server map services use schema locking to prevent other users from altering the geodatabase schema while the service is running. If the schema locks are impeding your workflow, you can disable them by manually adding the SchemaLockingEnabled tag to your service configuration file and setting it to false. For more information about editing the service configuration file and how to use this tag, see Service configuration files.
Globe services cannot be published inside other globe services. It's likely that the missing layers come from other globe services that were in the ArcGlobe document (.3dd) that you tried to publish. For example, the default ArcGIS Online layers in ArcGlobe are globe services that are hosted by ESRI and will be removed when you publish a globe service.
If you want to see the missing layers, add your globe service to a new .3dd or .nmf document. Then you can connect to the appropriate servers (such as ArcGIS Online) and add the layers that were originally removed from your globe service.
There are several adjustments you can make to allow sending large datasets to a geoprocessing service.
For ArcGIS Server Internet services, make sure the maximum message request size is large enough. In your ArcGIS Instance directory (Example: C:\Inetpub\wwwroot\ArcGIS) open the services directory, find the web.config file, and edit the maxRequestLength in the httpRuntime tag. If the tag is not there, you'll need to add it:
<system.web>
...
<httpRuntime maxRequestLength="20000"></httpRuntime>
</system.web>
The default size in kilobytes of the maxRequestLength is 4096. You can increase this size as shown above to allow for larger requests.
When working with large geoprocessing jobs, you may also need to increase the maximum time a client can use a service. You can configure this on the Pooling tab of the Service Properties dialog box. The default is 600 seconds, or 10 minutes. See Tuning and configuring services to learn about the time-outs associated with a service.
A frequent reason the map appears blank is that it references a stopped, unavailable, or nonexistent map service. Network traffic, permissions issues, and modifications to service and file system properties can all lead to a blank map. The section "Administering the server" in this help system documents how to prepare for and avoid these issues.
Your map may also appear blank if your Web server machine has an underscore (_) character in its name. See ESRI Knowledge Base article 32503 for more information about this limitation.
This may be due to impersonation behavior with Windows 2000 that Microsoft has documented: http://support.microsoft.com/default.aspx?scid=kb;en-us;810204. You will need to give the ASPNET account "Act as part of the Operating System" privileges for your applications to work. To do this, follow these steps:
When running a Geoprocessing task inside a Web application, you have the option to save a file of information about the results, which you can use in future application sessions. The place where the file is saved is determined by the VirtualDirectory property of the Geoprocessing task. At run time, if the account that runs the Web application does not have sufficient access to this directory, you will see the Access Denied error message. A workaround is to open the application in Visual Studio, right-click the name of the project in the Solution Explorer, then click Add ArcGIS Identity. Through ASP.NET impersonation, the user name and password you specify will be used to run the application. This user name and password should have Read and Write permissions to the VirtualDirectory of the Geoprocessing task.
These links will only be available if the VirtualDirectory for the Geoprocessing task has execute permissions set to "Scripts only" in IIS Manager.
Learn more about configuring the VirtualDirectory for the Geoprocessing task
Searches against an ArcGIS Server map service can sometimes return more records than expected. This is because of the way the search works. When you define the Search Attributes task in Manager, you specify the layers to search and the fields in those layers to search. Although it appears in Manager that specific fields will be searched for specific layers, in reality you are simply defining a list of layers to search and a list of fields to search in. Thus, if two layers have the same field name, the field will be searched in both layers.
If the Search Task is configured to work with two layers that have the same name, and a match is found in both, the application may return an error. This is a known issue.
When you run a task, each result feature is normally listed next to a check box that can be used to highlight the feature. Also, you can right-click on a result and get options to Zoom To or Pan To the selected feature.
The check boxes may be missing, and therefore the feature cannot be selected. This happens when the Shape field of the dataset has been marked as hidden in the map document. The Web application cannot get the geometry that it needs, and the check boxes are omitted.
When the dataset's Shape field is hidden, results from a Query task don't have check boxes, the selected feature is not highlighted on the map, and the Zoom To and Pan To options are not available.
In the case of results from a Search or Identify, the check boxes appear but the selected feature cannot be highlighted on the map.
To allow features to be selected and navigated to:
Image services are new at ArcGIS Server 9.3 and represent raster datasets or Image Server compiled image service definitions that have been made available through ArcGIS Server. The workflow for using an image service inside of a Web ADF application is to enable the WMS capability on the image service, then add the service to the application as WMS. WMS is an open specification for making map images available over the Web.
When viewing the image service properties, or when creating the image service, you will see a list of checkboxes of capabilities that you can enable. You should check WMS and note the URL of the service that will be created. Manager and the Web ADF controls support adding WMS services to your web applications. You'll just need to enter the URL.
See WMS services and Selecting services to display to learn more about WMS services and how to add them to Web applications in Manager.
To make sure your Web application is optimized for performance, review the topic Performance tips for Web applications. ESRI is constantly researching ways to improve performance in the core software and ways that you can improve the performance of existing applications by adjusting hardware and software settings. Best practices are continually added to the Web Help in the topic linked above.
If your GIS server is failing or experiencing severe performance degradation under a heavy load (over 25 concurrent requests per second) it may be that the Local Security Authority Subsystem Service (lsass.exe) has become overworked. ESRI Knowledge Base Articles 32620 (Windows Server 2003) or 32622 (Windows XP) explain how to adjust your server to work around this situation.