|
|
Sometimes things go wrong. Components often display confusing and unexpected error messages, or behave outside their normal parameters. Identification of these symptoms, their possible causes, and their possible solutions could potentially be a time-consuming and frustrating experience.
Ideally, the material in this appendix will provide a framework to help you identify and resolve routine problems. When problems are not so routine, this appendix will also help you by ruling out possibilities before you call the Cisco Technical Assistance Center. This will ensure a faster and more beneficial resolution to your problem.
This appendix includes the following sections:
This section includes the following topics:
Symptom You see the following error message:
Cannot write message to Director, errno =2
Possible Cause This error occurs when smid tries to write to a socket that does not exist. This may occur because the Director's nrdirmap application has not created its communication socket in /usr/nr/tmp because the OpenView user interface (ovw) was not started.
Recommended Action First ensure that the underlying NetRanger services, such as postofficed and smid are running by typing nrstatus at the command line. If the services are not running, type nrstart to manually start them. Then start openview by typing ovw & at the command line.
Symptom You see the following error message:
Cannot write message to Director, errno = 233
Possible Cause This error message is generated when smid writes to a socket whose buffer is overflowing. This can occur when the Director's nrdirmap application is not running.
Recommended Action Ensure that the OpenView user interface (ovw) is running by executing ovw &. This will automatically start nrdirmap.
Symptom You see the following error message:
Cannot write message to Director, errno = 239
Possible Cause This error message occurs when smid and nrdirmap do not have adequate permissions to communicate via sockets in /usr/nr/tmp.
Recommended Action Ensure that the smid process is owned by user netrangr, and that nrdirmap runs as SUID netrangr.
Symptom One of the following error messages is generated by HP OpenView upon startup:
848967818198: WgcSema can't access `/usr/nr/etc/nrdirmap.semaphore' Error: nrdirmap:main, semaphore initialization failed.
Possible Cause The nrdirmap.semaphore file in /usr/nr/etc has been deleted or has improper permissions.
Recommended Action Ensure that /usr/nr/etc/nrdirmap.semaphore exists and that the netrangr user and group accounts have read access to the file. If the file does not exist, you can create the file with a text editor, such as vi. Add a single character to the file and save it in /usr/nr/etc with the name nrdirmap.semaphore. Also ensure that /usr/nr/bin/nrdirmap runs as suid netrangr.
Symptom You see the following error message:
nrdirmap: fatal: libovw.so.1: can't open file: errno=2
Possible Cause The LD_LIBRARY_PATH environment variable is not set properly in your user environment. This may indicate that you are logged on to the Director platform as the wrong user.
Recommended Action Follow the instructions in "Installation and Configuration," for setting up an HP OpenView environment for user accounts other than netrangr. If the user account is based on either the Bourne or Korn shell, the following lines should exist in the user's $HOME/.profile:
if [ -d /opt/OV ] ; then. /opt/OV/bin/ov.envvars.sh PATH=$OV_BIN:$PATH export PATH LD_LIBRARY_PATH=$OV_LIB:$LD_LIBRARY_PATH export LD_LIBRARY_PATH
fi
If the user must use a shell other than ksh, then the lines above must be translated into the appropriate scripting language and placed in the appropriate startup file.
This section includes the following topics:
Symptom The Director's security map contains a Sensor icon, but fails to show any events for that Sensor.
Possible Cause The Director's Severity Status attributes are set higher than the level of alarms being generated by the Sensor.
Recommended Action On the Director interface, highlight the icon for the Sensor system and then either press Ctrl-O or click Describe/Modify on the Edit menu. Then select NetRanger/Director and click View/Modify. Ensure that the Minimum Marginal and Minimum Critical status thresholds are low enough to register events from the Sensor in question.
Symptom The Director's security map contains a Sensor icon, but fails to show any events for that Sensor.
Possible Cause The level of alarms generated by the Sensor system fall below the routing threshold set in the /usr/nr/etc/destinations file.
Recommended Action If the Director Severity Status thresholds are set properly, ensure that the routing threshold in the Sensor's /usr/nr/etc/destinations file is set low enough to route information to the Director.
Symptom You see the following error message:
ApplicationAppId.HostId.OrgIdhas reached maximum number of alarms.
Possible Cause The application mentioned in the error message has 1000 alarm icons represented on its OpenView child submap. The Director will not create more than
1000 icons on a submap (window), because OpenView can behave unpredictably when this happens.
Recommended Action Delete the alarm icons on the crowded submap. The Director will resume creating alarm icons on the submap for any new events. In order to view iconic representations of the events that nrdirmap diverted to /usr/nr/var, delete the icons on the map, then shut down and restart the user interface.
Symptom A Sensor's alarms are properly displayed on the Director security map, but information on those alarms does not appear in the Show Current Events window and the event log file in the Director's /usr/nr/var directory does not contain any records from that Sensor.
Possible Cause The Director loggerd service is not listed as a destination in either the Sensor's or the Director's configuration files.
Recommended Action Use nrConfigure to create an entry in the Sensor's /usr/nr/etc/destinations file for the Director's loggerd service, or create a DupDestination entry in the Director's /usr/nr/etc/smid.conf file to redirect event data to loggerd from smid.
Symptom Information is properly displayed in the Director's Show Current Events window, but the cursor turns into an hourglass and never changes back.
Possible Cause The current events utility continues to pull information from the Director log files as long as the window is up.
Recommended Action Click Stop to terminate the filtering application. You can then use the scrollbars and menu options to look at the data. Click Close to exit this window.
Symptom Unable to access a Sensor or any of the NetRanger services running on the system.
Possible Cause The Sensor services are not running properly.
Recommended Action Telnet to the Sensor and run nrstop. Examine the error files in /usr/nr/var. Restart the Sensor by typing nrstart.
Symptom Unable to start or stop the NetRanger daemon processes when running nrstart or nrstop.
Possible Cause You are trying to run these utilities from an account that does not have access rights to the Sensor daemons.
Recommended Action Ensure you are logged on to the Sensor or Director systems under the same user account that was used to start its daemon services. (The default is user netrangr.)
This section includes the following topics:
Symptom Cannot determine if Oracle is installed.
Recommended Action Check local and mounted file systems using commands df, mount, and find. Look for oracle and product/.
Symptom Cannot determine if Oracle is running.
Recommended Action Run ps -ef | grep ora from the command line to check if Oracle is running.
Symptom Oracle Installer (orainst) was unable to find any products to install.
Possible Cause start.sh was not run prior to starting orainst.
Recommended Action Run /cdrom/cdrom0/orainst/start.sh to prepare your environment for the orainst program.
Symptom One of the following messages is displayed:
sqlplus: not found sqlldr: not found
Possible Cause The Oracle bin directory is not present or specified properly in your $PATH.
Recommended Action Set $PATH to include $ORACLE_HOME/bin.
Symptom The following error message is displayed when you try to run sqlplus:
~~~/oracle/product/7.3.2/bin/sqlplus: cannot open
Possible Cause The shell finds sqlplus, but it cannot be executed. This can occur when your $PATH includes references to the wrong versions of the Oracle binaries. For example, you have mounted the wrong Oracle directories from a file server. Therefore, you are trying to execute HPUX binaries on a SPARC system.
Recommended Action Ensure that the $ORACLE_HOME directory contains the proper binaries for the platform you are running. Refer to "Installing an Oracle RDBMS" in "RDBMS Reference."
Symptom sqlplus, sqlldr, or sapx fail with the following SID error message:
ERROR: ORA-01034: ORACLE not available ORA-07200: slsid: oracle_sid not set
Possible Cause The ORACLE_SID environment variable, which identifies which database instance to use, was not set properly prior to starting sqlplus.
Recommended Action Set the ORACLE_SID environment variable to the name of your database instance. You can find out your database instance name by running ps -ef | grep ora. The string after the last underbar in the returned text is the database instance name.
Symptom sqlplus, sqlldr, or sapx fails with the following libc error message:
libc.so.xxx: can't do something
Possible Cause $ORACLE_HOME/lib is not part of the LD_LIBRARY_PATH environment variable.
Recommended Action Add ORACLE_HOME/lib to the LD_LIBRARY_PATH environment variable. If you are running either the Bourne or Korn shell ensure that your $HOME/.profile contains the following entries:
LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$ORACLE_HOME/lib export LD_LIBRARAY_PATH
Symptom sqlplus, sqlldr, or sapx fail with the following TNS error message:
ERROR: ORA-12154: TNS: could not resolve service name
Possible Cause Oracle cannot understand the name specified in your connect string.
Recommended Action Ensure that the Oracle file tnsnames.ora resides in its proper location (usually $ORACLE_HOME/admin/network) and that it is properly formatted. Then use the tnsping utility to test sqlnet connectivity to your remote database.
Symptom sqlplus, sqlldr, or sapx return a TNS or USER/PASSWORD error message.
Possible Cause Improper connect string.
Recommended Action Correct the syntax on the connect string. If specifying the password on the command line, type sqlplus user/password@host. Otherwise, type sqlplus user@host and sqlplus will prompt for a password.
This section includes the following topics:
Symptom Instrumentation shows proper configuration, but no actions are being performed.
Possible Cause If there are no FM_Action items in sapd.conf, then the install procedure did not properly copy files into sapd.conf due to upgrade from NetRanger 1.2.x to 1.3.x.
Recommended Action Manually copy the sapd.conf.nsx or sapd.conf.director from /usr/nr/etc/wgc/templates into /usr/nr/etc/, replacing the sapd.conf file.
Symptom There are extraneous files in the /usr/nr/var directory structure, and there exists a /usr/nr/var/old directory.
Possible Cause You have upgraded from NetRanger 1.2.x to 1.3.x, which does away with the /usr/nr/var/old directory. The upgrade has also left behind many files in /usr/nr/var.
Recommended Action Clean the extraneous files, delete the /usr/nr/var/old directory, and archive the /usr/nr/var/dump directory.
Symptom SQL queries do not display data.
Possible Cause You did not enter % as a wildcard.
Recommended Action Use % as a wildcard in your SQL queries.
Symptom From the SQLPLUS prompt, entering @event1, @space1, @time1, or @system1 does not return proper data.
Possible Cause You are using the wrong command line parameter.
Recommended Action SAP 1.3.x requires that you enter either @event, @space, @time, or @system. You will be prompted for the desired drill-down level (e.g., 1, 2, 3).
Symptom Queries do not display new signatures.
Possible Cause You have upgraded from NetRanger 1.2.x to 1.3.x without updating the nr_sigs and nr_orgs tables.
Recommended Action Update the nr_sigs table with /usr/nr/etc/signatures and the nr_orgs table with /usr/nr/etc/organizations. Use the code at the end of nrdb_master_create for this purpose.
Symptom Instrumentation shows a successful notify, but no mail notification has been sent.
Possible Cause mailx feature not invokable through the command line interface.
Recommended Action To ensure that mail can be sent from a command line, from the command line, use mailx to send mail to yourself. If mail is not sent, set the domainname by typing domainname your_domain_name from the command line (on Solaris, you can add the name of your domain to the /etc/defaultdomain file). Then add your mail server information to /etc/hosts, with the following format: IP_address server_name mailhost where IP_address is the IP address of your mail server, and server_name is its DNS server name.
Symptom The sapx database loader fails with a JDBC-related error message (ora-1461).
Possible Cause You are using the NT Oracle 8 database server.
Recommended Action You have three options for bypassing this error:
1. Bypass the default sapx loader by using the alternate loading templates in /usr/nr/bin/sap/sql/skel.
2. Use a UNIX Ora7 or Ora8 server. Cisco has successfully tested the server software on Solaris Sparc, x86, HP-UX, and AIX.
3. Upgrade NT Ora 8.0.4.0.0 to 8.0.4.0.4. This upgrade should solve the JDBC problems, but has not been tested.
This section includes the following topics:
Symptom During initial startup, nrConfigure will cause a core dump.
Recommended Action Restart nrConfigure. nrConfigure will then restart without error.
Symptom HP-UX versions of nrConfigure run slowly on some HP machines.
Possible Cause There is not a reliable JIT Java Compiler on HP-UX. This performance problem can sometimes lead to confusion when the user interface is sluggish and users initiate actions several times because of poor performance. For example, rapid successive mouse clicks can lead to unexpected behavior by nrConfigure. Another case involves Java errors scrolling on console and the Java application screens crashing.
Recommended Action Retry your previous steps. In most cases, a second or third attempt is successful.
Symptom Launching the NSDB or online help launches a new copy of the HTML browser, instead of refreshing the existing HTML browser window.
Recommended Action If you use Netscape, you can configure NetRanger to load all HTML pages into a single browser window. To do this, follow these steps:
Step 1 Open the /usr/nr/etc/nrConfigure.conf file in a text editor.
Step 2 Change the value of the Browser token to the following value:
Browser=/usr/nr/bin/director/nrSingleBrowser
Step 3 Change the value of the NetscapeLocation token to the following value:
NetscapeLocation=/opt/netscape/netscape
Step 4 Save your changes and close the editing session.
Posted: Fri Jul 28 08:37:59 PDT 2000
Copyright 1989-2000©Cisco Systems Inc.