Administration > Data Collector Administration > SiteScope Administration > External Integrations and Functionality > Mercury Self-Alert Monitor > Troubleshooting Directory and Log File Errors
Troubleshooting Directory and Log File Errors
If one or all Directory monitors indicate a "directory not found" error, and all log file monitors indicate an "unable to read log file" error. These errors can happen when you monitor a Mercury Business Availability Center system running on Windows operating systems. The cause of this type of error is likely because the Mercury Self-Alert Monitor assumes that the Mercury Business Availability Center installation path is: C:\MercuryAM.
Note: When making any of the changes described here, you should make a backup of the files that are to be modified.
Note: After making any of the changes described here, you must stop and restart the SiteScope service for the changes to take effect.
To resolve this problem, do one (or more) of the following:
Modify the Templates
Use the following steps to modify the templates.
To modify the templates:
- Edit the Mercury Self-Alert Monitor monitor templates in the <SiteScope root directory>templates.sets.topazWatchdog directory, and replace the $TOPAZ_FOLDER$ string with the location on the local disk where Mercury Business Availability Center is installed. For example, if Mercury Business Availability Center is installed on volume D, change $TOPAZ_FOLDER$ to D$\Topaz.
If the Mercury Business Availability Center folder is shared by the name MercuryAM, change $TOPAZ_FOLDER$ to be MercuryAM.
- Save the changes to the file.
- Update the Self-Alert Monitor using the following steps:
Modify the Configuration File
Use the following steps to modify the configuration file.
To modify the configuration file:
- Open <SiteScope root directory>\groups\watchdog.config, and change the value of the _twdTopazFolder parameter.
For example, you can set this parameter to D$:\MercuryAM by changing the line
_twdTopazFolder=C$ \MercuryAM
to:
_twdTopazFolder=D$ \MercuryAM.
You can set this parameter a network drive share alias name by changing the line
_twdTopazFolder=C$ \MercuryAM
to:
_twdTopazFolder=<NETWORK SHARE ALIAS>.- Save the changes to the file.
- Update the Self-Alert Monitor using the following steps:
Note: If you clear the Mercury Self-Alert Monitor settings, the _twdTopazFolder parameter is reset to C$ \Topaz.
Edit the Individual Monitors
Edit each problematic monitor, and change its directory path or the log file pathname attributes to the correct directory or file path.
After making any changes, stop and restart SiteScope.
More Troubleshooting Issues
The following table lists a number of other conditions that may occur when using the Mercury Self-Alert Monitor with Mercury Business Availability Center.
Problem Solution(s) In a Windows 2000 installation, drive letters are replaced by HarddiskVolume1, HarddiskVolume2, and so forth. For the solution, refer to: http://support.microsoft.com/support/kb/ articles/Q274/3/11.asp. (Microsoft Knowledge Base article number Q274311) In a Windows 2000 or 2003 installation, no disks are monitored by SiteScope. This occurs when a Windows 2000 installation has disk monitoring disabled by default. To enable disk monitoring, open a command line window, and enter the command diskperf -y. After restarting the computer, the disks are added to SiteScope. If the Mercury Business Availability Center Admin server is installed on Apache, on a machine that automatically runs Microsoft IIS, SiteScope may not measure the correct process, and performance may be impeded. Make sure that only the Web server that Mercury Business Availability Center uses is running. Some systems, for example, Windows 2000, have IIS automatically installed and running. Therefore, if Mercury Business Availability Center is installed on Apache, on a Windows 2000 server, both IIS and Apache will be running, which slows down performance and scalability.Stop the IIS service and disable it from running automatically. This will ensure that the machine runs faster, and that SiteScope measures the correct process. If a network drive to a server has been mapped with non-administrative credentials, you cannot open a new authenticated network connection with administrative permissions, or change the existing one. This is a Windows networking limitation. The solution is not to disconnect the mapped drive, as the connection will remain alive. The problem also occurs if you run Terminal services to a machine or just explore it with Explorer. It can be destroyed only after a reboot of the SiteScope machine. If you are still not able to monitor a server even if you know the administrator user name and password, you should run netstat -a, to see if there is an active (established) connection to that machine. SiteScope cannot connect to a remote Windows server. If a connection cannot be made, check the user access permissions that have been granted to the SiteScope account on the remote server. SiteScope requires remote registry permissions to be able to monitor server statistics. Try connecting to the remote machine using Perfmon. SiteScope is not allowed to use the permissions of a full administrator account. This may be for security reasons. SiteScope can be granted restricted monitoring access by editing certain Registry Keys. See the Enabling Non-Admin Users to Remotely Monitor with PERFMON support note on the Microsoft support site for more information. SiteScope cannot monitor a stand-alone server, or one that is part of a domain already visible to the SiteScope server. Try entering the machine name followed by a slash and the server name in the Login entry box. For example, type cats/administrator SiteScope does not monitor the load balancer machine. SiteScope cannot display data about load balancer machines because they are not Mercury Business Availability Center hosts.You can set up a standard SiteScope monitor to monitor the load balancer machines, and to send an alert when the load balancer is in error. For details, see Working with SiteScope Monitors in Configuring SiteScope Monitors. Monitors are returning errors on specific machines. Check whether the machine has been added to the Remote Windows Servers or Remote UNIX Servers list. For details, see "Defining Permissions for NT Servers" and "Defining Permissions for UNIX Servers".Check whether remoteRegistryService is running on the Mercury Business Availability Center machine.A Microsoft Windows problem causes the registry service RemoteRegistryService to hold too many handles that are not released. Every time SiteScope logs in to the Mercury Business Availability Center server, another handle is added. To see how many handles are being held by a process, display the Processes tab in the Windows Task Manager, and look for the Handles column. (If the Handles column does not appear, choose View > Select Columns > Handles.)To release handles, restart the service, either manually:
net stop "remote registry"
net start "remote registry"
or with a script that is set to run, for example, every six hours (but can depend on the specific setup), by using Windows Task Scheduler (Scheduled Tasks in Explorer).Example of the script:
at 12:00 /every:Su,M,T,W,TH,F,S net stop RemoteRegistry at 12:01 /every:Su,M,T,W,TH,F,S net start RemoteRegistry at 18:00 /every:Su,M,T,W,TH,F,S net stop RemoteRegistry at 18:01 /every:Su,M,T,W,TH,F,S net start RemoteRegistry at 00:00 /every:Su,M,T,W,TH,F,S net stop RemoteRegistry at 00:01 /every:Su,M,T,W,TH,F,S net start RemoteRegistry at 06:00 /every:Su,M,T,W,TH,F,S net stop RemoteRegistry at 06:01 /every:Su,M,T,W,TH,F,S net start RemoteRegistry Sometimes counters for perfmon objects are disabled on the Mercury Business Availability Center machine (for example, for the Process object). To monitor such a machine, you must enable these objects. You can do this either with the help of the Windows Resources Kit, or via the registry.With the Windows Resources Kit, in the Extensible Counter Kit dialogue box, check that the Performance Counters Enabled check box is selected, for the PerfProc and PerfOS objects. Using the registry, check that the following files are set to 0:
HKEY_LOCAL_MACHINE\SYSTEM\
CurrentControlSet\Services\PerfProc\
Performance]"Disable Performance Counters"= dword:00000000
and
HKEY_LOCAL_MACHINE\SYSTEM\
CurrentControlSet\Services\PerfOS\
Performance]"Disable Performance Counters"= dword:00000000 SiteScope sends false alerts. This is an integration issue, with two causes:SiteScope has been disconnected from Mercury Business Availability Center, and alerts are sent several minutes after the disconnection. To prevent this, clear the check box next to the SiteScope machines, or use the Data Collector Maintenance page, accessed from Admin > Platform > Data Collection > Data Collector Maintenance in Mercury Business Availability Center, to remove the SiteScope no longer in use.SiteScope has been moved from one profile to another, and alerts are sent during the period when SiteScope is in downtime. To prevent false alerts, delete the old SiteScope profile (preferred), or disable the alerts for this host. The same host appears twice in the SiteScope main panel. During probe definition, the probe's location is registered by its relationship to the location of the Business Process Monitor host that is running the probe. If the definition of this Business Process Monitor is changed, and if its new location is different from the previous one, the new registration and location for the probe is added to the Host table in the Mercury Business Availability Center management database. The result is that the table includes two probe hosts with the same name but with different locations. This causes SiteScope to display the probe host twice in the main panel.
| Copyright 2006 Mercury Interactive Corporation |