For the Windows NT® 4.0 Operating System
Listed below are solutions and workarounds to problems that have been encountered while using Network Node Manager 6.0 (NNM).
Please also visit the Release Notes addendum at http://ovweb.external.hp.com/nnm/NNM6.0/relNoteUpd/relNoteUpdate.htm for any more recent updates or workarounds.
There is also a troubleshooting section below.
Below are the known problems and workarounds for the following modules:
Be sure to re-install your latest NT service pack after installing any Microsoft components or you will get messages about mismatched ODBC DLLs or the SNMP extension agent will have an exception.
Remember to reinstall your latest service pack after installing an Microsoft components, such as the SNMP agent, Peer Web Services, or IIS. Not reinstalling service packs can cause the Microsoft SNMP agent to throw an exception or creates ODBC mismatched DLL messages.
Use the forms in the $OV_CONF/OVLicense/nnm/forms/C directory to make password requests for the "HP OpenView ECS Designer for NNM and ITO" product. The forms in the $OV_CONF/ecs/forms/C directory are for the standalone ECS Designer product only and should not be used.
You can now access NNM running on a management station from remote computers over the WWW using a
JDK 1.1 Compliant Web Browser.| Operating System | Browser | Browser Information |
|---|---|---|
| Solaris 2.5.1 | Use Netscape browser for best results | |
| Netscape 4.06 | Netscape may crash when first brought up. See Netscape README file for required operating system patches. | |
| Internet Explorer 4.01 | UNIX Microsoft Internet Explorer displays error when launching NNM Java
applets. Microsoft Internet Explorer on Solaris --Menus not displaying properly. A possible workaround is to run /usr/openwin/bin/xlsfonts -o utopia-regular >/dev/null |
|
| Solaris 2.6 | Use Netscape browser for best results | |
| Internet Explorer 4.01 | MS Explorer on Solaris--Menus not displaying properly. A possible workaround is to run /usr/openwin/bin/xlsfonts -o utopia-regular >/dev/null | |
| Netscape 4.06 | Solaris 2.6 Netscape has corrupt local display with release notes loaded due to not rendering text fonts correctly. See the Netscape README for the two xset commands to work around this. | |
| Windows NT 4.0 | Use Internet Explorer browser for best results | |
| Internet Explorer 4.01 SP1 or Netscape 4.06 |
Java applications returning HTTP/1.0 500 Server Error and not loading. The workaround is to reinstall Windows NT Service Pack 3. | |
| Windows 95/Windows 98 | No browser has yet been fully qualified for these platforms | |
| HP-UX 10.20 | Use Netscape browser for any results | |
| Netscape 4.06 | <NONE> | |
| There are many incompatibility issues between NNM and Microsoft Internet Explorer for HP-UX. | ||
| HP-UX 11.0 | Use Netscape browser for any results | |
| Netscape 4.06 for HP-UX 10.20 | HP-UX 11.0 is not officially supported by Netscape but seems to work on 32 bit hardware. | |
| I | Internet Explorer 4.01 (pre-release) | There are many incompatibility issues between NNM and Microsoft Internet Explorer for HP-UX. |
If are still having problems starting the launcher, you may have a configuration problem with your web browser.
If you try to bring up more than one instance of Netscape on Solaris or HP-UX, you may get a dialog box stating:
"Netscape has detected a /home/user/.netscape/lock file.
This may indicate that another user is running Netscape."
Please ignore this message.
Do Not Close Launcher Window
Do NOT close the Launcher window, itself, when launching applications. Some applications use the frames within the Launcher window for executing web scripts. If the Launcher window is closed, it will cause the application that was launched from the window to close. This is particularly true when launching the HP OpenView Network Presenter. The Launcher window may be iconized to remove its presence instead of being closed.
You must have NNM running on the management station with the desired map displayed in order for the web interface to operate.
Using Internet Explorer 4.01 or greater:
Behavior is variable when doing a right click on a symbol in the scoping pane or tabular
view of the data in the content area to bring up the popup menu. It is dependent upon the
interaction of the browser and the third party component we use to implement the scoping
pane and tabular view. With Internet Explorer running on Windows NT, the selections may
change.
Using Netscape 4.06 or greater:
Behavior is variable when doing a right click on a symbol in the scoping pane or tabular
view of the data in the content area to bring up the popup menu. It is dependent upon the
interaction of the browser and the third party component we use to implement the
scoping pane and tabular view. With Netscape on UNIX, unselect anything selected before
doing a right click. With Netscape running on Windows NT, the selections may change.
With the Netscape 4.06 browser running on HPUX, the Network Presenter's context button can become corrupted if a symbol on the root submap is deleted. The correct number of symbols will be displayed but the symbols no longer represent the correct context. Choosing one of the symbols will result in the incorrect context being displayed. (NSMfc13446)
If using a proxy server, and the DNS server running on the proxy server cannot resolve an NNM data presenter hostname, the graphic lines for the tree structure cannot be located, and will appear as generic icon symbols. If this happens, disable use of the proxy server, clear the browser caches, and reload the page. If the problem persists, modify the http configuration file on the http server to provide the fully qualified name of the http server:
Using Internet Explorer 4.01 or greater:
The JavaScript-based tree used in the contributed Report Presenter and SNMP Data Viewer
may cause the user to have to re-position the scroll bar after opening a folder (or
selecting a link). Whenever the screen is refreshed after opening a folder (or selecting
an item) the scroll bar is re-positioned at the top. When users are working from the
bottom of the list they will have to re-position the scroll back to where it had been
prior to their selection. The basic problem lies in the implementation of JavaScript in
Internet Explorer and may be corrected in a future release.
Using Netscape 4.06 or greater:
Resizing the HP OpenView SNMP data presenter web pages may result in certain frame data
not being redisplayed. Reload the page (e.g., push the Reload button, or right-click and
select Reload) to correct the problem.
When the Web Alarm Browser is connecting to an NNM management station running on Windows NT, printing only works with a local printer connected to that Windows NT based management station. The printer must be configured on one of the following ports LPT1(PRN), LPT2, or LPT3.
Using Internet Explorer 4.01 or greater:
There is a problem with menus where a shadow of a "Warning Applet Window"
appears under every second level menu, such as Actions->Acknowledge, Actions->UnAcknowledge,
or Actions->Delete.
Netscape:
We have found problems while running the ECS configuration management GUI while using
Netscape Enterprise Server. The problems are related to file locking. The default server
we ship with NNM does not suffer from these problems.
Netscape does not yet officially support HPUX 11.X with it's Navigator product, while using the HPUX 10.X version of Navigator running on HPUX 11.X, when you try to to add a row in several parameter dialogs, the vertical slide bar does not work or appear until you resize the window.
NNM 6.01 currently only supports the following web servers:
You may replace the NNM Web Server with your own web server by following the instructions on the CD in doc/WhitePapers/changeWebServer.txt
ERROR: "Cannot look up field id for name field IP Address" appears in the Alarm Browser (NSMfc11943)
If ovtopmd is started after clearing the databases in $OV_DB/openview, this error message will appear in the alarm browser: This is a harmless error message which can be avoided in the future by performing the following steps after clearing out the NNM databases:
ovstart ovwdb
ovw -fieldsERROR: ipmap reports number of licensed connectors as -1 (NSMfc10339)
If you select the Internet symbol and look at the Object Properties ->IPmap there is an entry for the number of licensed connectors. When the unlimited license is installed this field will contain the value "-1". When any other license is installed, the number of licensed connectors will appear: i.e. 250 for the instant on license. You can safely ignore this message.
ERROR: You remove a symbol from the map, yet it remains on the map. (NSMfc12765)
Timing conditions while manually adding 250th node to a system licensed to have only 250 managed nodes while discovery is active may cause a 'removed symbol' to remain on the map. This results from both the user and netmon attempting to add the last managed node, but in this case netmon wins. To prevent the problem from happening, turn off discovery using "Options->Network Polling Configuration" menu item, delete or unmanage nodes to bring the managed node count below 250, manually add the desired nodes, and re-enable discovery. To clean up the map, run ovtopofix (see the ovtopofix man/reference page for details)
New Non-Web Alarm Browser (xnmevents) Behavior
All operator actions, Acknowledge, UnAcknowledge, Delete, Assign Severity, Assign Category, are global (affecting all browsers).Alarm Browser (xnmevents) no longer maintains per user state files. There is only one common state file. Old per user state files are not compatible with NNM 6.01 and will be ignored.
Some correlated events, from the PairWise correlation, will cause the Alarm Browser to redraw twice on Windows NT. In the rare situations where this occurs, this could take up to 16 seconds of time. (NSMfc12850)
When using databases other than the embedded database, the user needs to manually instantiate the data warehouse schemas using ovdwsetup.ovpl before using the data warehouse.
All export tools are enabled for ODBC. To use Oracle, sql*net must be
installed and configured. A default datasource has been configured in the /etc/opt/OV/share/conf/analysis/system_odbc.ini
file, OVoracle. To switch over, change the "ServerName" parameter
in that file to your server as defined in tnsnames.ora.
Then run:
"ovdwsetup.ovpl -rdb OVoracle -type oracle -u <DBuser> -password <pass>", where <DBuser> is your oracle user and <pass> is that user's password. Oracle user OSauthentication is no longer supported.
During our testing, we have experienced situations where killing or interrupting the database client program can have an adverse effect on some database servers.While this is still being investigated it appears that some versions of Microsoft SQL Server will fail when the client is canceled via "Control-C".
If you are using the NNM embedded database technology, you may also experience a behavior where for a short time after interruption, the database server may think the client is still active. This could result in "primary key contraint violations" or some type of concurrency problem, if the same command is immediately restarted. However this should clear up within 30 seconds and future access to the database will not be effected.
Clicking on the Stop button in the "Tools/Data Warehouse/Export Trend Data" dialog window will not interrupt the trend export process. It will finish to completion.
You must reinstall your Windows NT service pack after adding any Microsoft Components such as Microsoft Peer Web Services, or you will see the following error (or an SNMP error).
ODBC Version Mismatch dialog:
You may get an "ODBC Driver Manager" popup window for a version mismatch for
various ODBC utilities such as:
The ODBC resource DLL
(C:\WINNT40\System32\odbcint.dll) is a different
version than the ODBC driver manager
(C:\WINNNT40\System32\ODBC32.dll).
You need to reinstall the ODBC components to ensure proper operation. This happens if you install *any* application which supplies ODBC components, not necessarily just NNM. The only way to prevent this popup is to reinstall your last Service Pack. NNM does NOT include any Windows NT ODBC components. This mismatched DLL problem is a known issue with Windows NT. Further information can be retrieved from Microsoft Article PSS ID Number: Q170769.
If you are using ovbackup.ovpl and are using the NNM Data Warehouse, when the ovbackup.ovpl command is executed, it will also instruct the Embedded Database server to initiate the online backup. In order to allow a roll-forward recovery the default backup schedule must be deactivated. The process for doing this is:
- copy the $OV_DB/analysis/default/solid.ini file to $OV_DB/analysis/default/solid.ini.old file.
- Edit the $OV_DB/analysis/default/solid.ini file.
- Comment out the "At=<time> backup" entry, by inserting a ";" at the beginning of the line. An example would be:
;At=01:00 backup- Save the $OV_DB/analysis/default/solid.ini file.
The embedded database will now be backed up only when the ovbackup.ovpl command is run. If use of ovbackup.ovpl is stopped at a future this, the default backup should be reinstated by copying the $OV_DB/analysis/default/solid.ini.old back to $OV_DB/analysis/default/solid.ini.
For Japanese NT Remote Consoles connecting to HP-UX or Solaris Management Stations, the 3rd party NT NFS product must be able to resolve the Japanese_Japan.932 symbolic links in the registration, symbols, and conf subdirectories. If the ovw menus and symbols on the Remote Console are in English, this means that the symbolic links are not working correctly. For the third party NFS product Disk Access, you need to enable symbolic links from the Disk Access applet in the Control Panel, not from the Administrator Utility. The default for Disk Access is to not enable symbolic links.
No known problems on Windows NT.
No known problems on Windows NT.
(NSMfc13038)
The following procedure will cause the Event Configuration window to hang:
- start the Event Configuration Window from NNM's user interface, select:
Options->Event Configuration- pause NNM for back up (ovpause portion of the backup script)
- Edit an the event source using "Add From Map":
Edit->Add->Event->Add From MapThe workaround is to not perform "Add From Map" during the paused portion of NNM's backup procedure. Or, if it's already paused and you cannot wait until backup has finished, from the command prompt:
- Windows NT:
Run \OpenView\bin\ovresume- UNIX:
Run $OV_BIN/ovresume
NNM management station running on Windows NT (interaction with Microsoft's SNMP service)
When you install NNM, the Microsoft SNMP Agent will be effectively disabled by causing it to listen for incoming SNMP requests at UDP port 8161 instead of port 161. During installation, NNM modifies the %SystemRoot%\system32\etc\services file. This will allow Microsoft's snmp.exe to continue to run, and to allow other applications which depend on the Microsoft agent to properly install, but no SNMP requests will be directed to Microsoft's SNMP service.
The SNMP Research Emanate SNMP Agent (snmpdm.exe) and adapter (wpaagt.exe) will instead be used to process the SNMP requests. Microsoft SNMP Service installation is required so that the Microsoft extension agents are installed (all SNMP agent extension DLL's are still used, only the Microsoft main agent is not used).
NOTE: All information entered in the "Microsoft SNMP Properties" dialog will be ignored. All configuration MUST be done through the \OpenView\conf\SNMPAgent\snmpd.conf file. This information includes SNMP trap destinations and community names.
HP OpenView Launcher and Network Presenter utilize the operating system's "iconv" libraries to convert data stored in the OV Launcher registration files and the OVW registration files into Unicode. This iconv library on Solaris 2.6 has a defect where it incorrectly converts the ASCII "~" and "\" characters when iconv is applied to Japanese data (Shift-JIS or EUC). This defect has been corrected by Sun Microsystems via Solaris patch ID "106613-01". You may need to install patch 106613-01 to your Solaris 2.6 if any of your Japanese (Shift-JIS or EUC) registration files (for OV Launcher or OVW), installed on Solaris 2.6, contain ASCII "~" or "\" characters.
The information below is needed for all users when using NNM's Java-based programs in Japanese language mode on all platforms.
(NSMfc12758)
The HP OpenView Launcher and Network Presenter Java-based NNM features are not supported
on Netscape 4.06 on HP-UX, Solaris, and NT in Japanese for both Shift-JIS and EUC locales.
This is due to a Netscape defect, bug ID 316500. Until this defect is fixed, these two
tools will not be supported on Netscape under Japanese. The Java-based Alarm browser
and ECS Configuration Management tool are unaffected by this defect and hence, can be used
in Japanese mode under Netscape. However, they cannot be started from the Launcher. To
start up the Web-based Alarm Browser without using the Launcher, enter one of the
following URLs:
http://<MACHINE>/OvCgi/ovalarm.exe (to access an NNM
management station running on Windows NT)
http://<MACHINE>:8880/OvCgi/ovalarm.exe (to access an NNM
management station running on UNIX)
To start up ECS Configuration Management tool without using the Launcher, enter one of
the following URLs:
http://<MACHINE>/OvDocs/C/ecs/ecscmg.html (to access an NNM
management station running on Windows NT)
http://<MACHINE>:8880/OvDocs/C/ecs/ecscmg.html (to access an
NNM management station running on UNIX)
It is hoped that a future version of Netscape will allow support of these tools. Please
see the following URL for any updated support status of Netscape under Japanese:
http://ovweb.external.hp.com/nnm/NNM6.0/relNoteUpd/relNoteUpdate.htm
When the ECS Configuration Management tool is launched in Japanese on NT using any WEB browser, the resulting display will be partially English and partially Japanese. The correlation specific information is displayed in English. However, the tool-specific information (correlation independent information) is displayed in Japanese. This problem will be corrected in a future patch. In the interim, there is no workaround.
ja locale on Japanese Solaris 2.X NNM supports both ja and japanese locale for EUC codeset on Solaris. However,
when LANG is set to ja, NNM tools will be partially English and partially Japanese.
To correct this problem, do the following as root:
cd /opt/OV/lib/nls/ja
rm japanese
mv * ../japanese/.
cd /opt/OV/lib/nls
rmdir ja
ln -s /opt/OV/lib/nls/japanese /opt/OV/lib/nls/ja
Stop ovw, and restart NNM processes:
ovstop
ovstart
ovw
There are a number fields in the Data Warehouse that contain localized data. These fields are:
- the "text" field in the nnm_event_cat table
- the "text" field in the nnm_event_sev table
- the "message" field in the nnm_event_detail table
- the "high_time" field in the nnm_event_thresh table
- and the "low_time" field in the nnm_event_thresh table
As a result of the localized data in the nnm_event_cat table, the data in the "Alarm Category" column of the Alarms By Category report will be localized. In addition, the data in the "category" column of the "Alarms By Severity" detail report will be localized. As a result of the localized data in the nnm_event_sev table, the data in the "Alarm Severity" column of the Alarms By Severity report will be localized. In addition, the data in the "Severity" column of the "Alarms By Category" detail report will be localized. As a result of the localized data in the nnm_event_detail table, certain text in the message fields of both the Alarms By Severity and the Alarms By Category will be localized.
The "high_time" and "low_time" fields from the nnm_event_thresh table are not used in any of the reports provide with NNM 6.01. These fields are localized because they are populated from a localized version of the trapd.conf file by ovdwevent. The version of trapd.conf that is used as a data source is determined by the local set when ovdwevent is run.
Localized event data will be written to the Data Warehouse using the code set specified at the time ovdwevent is run. Attempts to view data using a code set other then the set used to export the data, can cause data to be displayed using unreadable characters. The most common cause of unreadable characters in reports generated from Data Warehouse data involves the use of Microsoft Excel. Microsoft NT 4.0 only supports the Shift-JIS and Unicode code sets. Therefore, an attempt to generate reports on NT using data exported to the Data Warehouse in a code set other than Shift-JIS or Unicode, will result in those reports containing unreadable characters.
Netscape offers a switch called -visual to use a specific server visual. This allows the user having a better color usage of the more expensive graphics cards. With this switch you may have netscape using a different set of 256 colors. There is a defect in Netscape 4.5 that if it is run in the non-default visual, Java applications are not displayed. They seem to run somewhere in the back plane. How can you tell how many visuals you have and which is the default?
run /usr/contrib/bin/X11/xdpyinfo ......
number of visuals: 3
default visual id: 0x21
visual:
visual id: 0x20
class: PseudoColor
depth: 8 planes
available colormap entries: 256
red, green, blue masks: 0x0, 0x0, 0x0
significant bits in color specification: 8 bits
visual:
visual id: 0x21
class: PseudoColor
depth: 8 planes
available colormap entries: 256
red, green, blue masks: 0x0, 0x0, 0x0
significant bits in color specification: 8 bits
visual:
visual id: 0x22
class: TrueColor
depth: 8 planes
available colormap entries: 8 per subfield
red, green, blue masks: 0xe0, 0x1c, 0x3
significant bits in color specification: 8 bits
netscape -visual 0x21
TIP: Look in the following directories for error log files and trace output. Look for the keyword "ERROR":
See Managing Your Network with NNM for information about starting with a well-configured network for best results.
If during installation NNM complains that your SNMP Agent needs to be updated, you may obtain the latest SNMP Agent from ftp://ovweb.external.hp.com/pub/NNM601
->NNM will automatically start up and begin discovering and displaying your network within five minutes of installation. However, if your discovery is not working (such as an empty internet icon on the top level map) or you would like to validate your installation, see Chapter 5 in Managing Your Network with NNM for information about troubleshooting the Discovery process. The following may help you isolate any problems:
If NNM fails to discover all the nodes you expect to find in your local network, individually ping first your local node and then the missing nodes to verify connectivity (using the Fault:Ping" menu item). Validate that your node "knows" about these by selecting your management station icon and selecting the Configuration:Network Configuration->ARP Cache menu item. You should then see these new nodes in the ARP cache. Select your management station and select the Fault:Network Connectivity->Poll Node menu item to have the netmon process query your node again. The new nodes should then be added to the map.
If NNM fails to discover any nodes, you can manually seed discovery through the user interface or command line. See Chapter 5 of Managing Your Network with NNM and the netmon reference page in NNM's online help (or the UNIX manpage) for details.
This is done by adding a node at the internet level which supports SNMP. Even if
this node is not really a gateway, NNM will properly discover the device, create the
necessary networks, and place it in the appropriate network. Verify the node
supports SNMP by running:
rnetstat -I <nodename>
To manually add this node through NNM's user interface:
To manually add this node through a command line (you must know the correct subnet mask) run:
echo <IPADDR> <NODENAME> | loadhosts -V -v -m <SUBNETMASK>
where <IPADDR> is the IP address of the node to add, <NODENAME> is the name of the node, and <SUBNETMASK> is the subnet mask for the network. An example is:
echo 192.18.1.1 myrouter | loadhosts -V -v -m 255.255.255.0
You can also manually seed discovery by using a seedfile.
After seeding discovery, you should see the map start displaying nodes based upon the information received from the node which supports SNMP, or see a list of nodes when running the ovtopodump command.
Make sure the database is executing by running "ovstatus ovdbcheck".
You can verify that data exists in your data warehouse by using the ovdwquery command, such as:
- To test for topology data in your data warehouse:
echo 'select ip_hostname from nnm_nodes;' |ovdwquery
- To test for event data in your data warehouse:
echo 'select message from nnm_event_detail;' |ovdwquery
- To test for trend (snmpCollect) data in your data warehouse:
echo 'select descript_id from snmp_trend_dataset;' |ovdwquerySee Reporting and Data Analysis with NNM for more information.