The information needed for troubleshooting can be divided broadly into operating system information and HiRDB information. Operating system information is collected using operating system commands. HiRDB information is collected using HiRDB commands.

If you use support services, the information given in the table below might be needed. The table below provides information needed for troubleshooting. The priority for collecting information is divided into three categories: a performance problem, a no-response, and abnormal termination. There are seven priority levels, with 1 being the highest priority and 7 being the lowest.

Table C-1 Information needed for troubleshooting (OS)

No.	Collected information	Collection method	Prf.	NR	AT
1	syslogfile	Use an operating system function (command).	1	1	1
2	CPU utilization and device status	Use the OS's sar command. For details about this command, see the documentation for the operating system.	3	4	3
3	CPU activity or memory status for the process	Use the OS's top command. For details about this command, see the documentation for the operating system.	3	4	3
4	Virtual memory information	Use the OS's vmstat command. For details about this command, see the documentation for the operating system.	3	4	3
5	Network status information	Use the OS's netstat command. For details about this command, see the documentation for the operating system.	3	4	3

Table C-2 Information needed for troubleshooting (HiRDB)

No.	Collected information	Collection method	Prf.	NR	AT
1	HiRDB error information	Use the pdinfoget command. Collect the output information.	2	2	2
2	Error log files	Error log files are output to the directory $PDDIR/spool/errlog.	2	2	2
3	Command log file	Command log files are output to the directory $PDDIR/spool/cmdlog.	2	2	2
4	HiRDB system definition information	Save files in the directory $PDDIR/conf on a medium such as a DAT.	4	5	4
5	Table and index definition information	Save files describing definition SQL used when defining tables or indexes on a medium such as a DAT.	4	--	--
6	Performance trace information files	These files are output to a file under the directory $PDDIR/spool/pdprf. You can use the pd_prf_output_directory operand to change the performance trace output directory. If online performance degradation can be tolerated, we recommend that you collect information by setting the performance trace collection level to 0000007f (detailed level) so that the cause can be quickly identified. Instead of changing the value specified for the pd_prf_level operand, you can also use the pdprflevel command to change the collection level while HiRDB is active.	4	6	6
7	Statistical information	Use the statistical information output start and end commands (pdstbegin and pdstend) to collect the information listed below: Statistical information on system operations (sys) Recommended interval: 1 hour Statistical information on the global buffer pool (buf) Statistical information for HiRDB files on database manipulation (fil) This information is output to the directories $PDDIR/spool/pdstj1 and pdstj2. After collecting the statistical information, use the statistics analysis utility (pdstedit command) to edit the statistical information.	6	--	--
8	SQL object information#1	Use the pdobils -C -H -R command. Execute this command at 60-second intervals and use an appending redirect (>>) for output to the target file.	5	--	--
9	SQL object information#2	Use the pdobils command with an SQL object number specified in the -N option.	6	--	--
10	Information about executing SQL statements#3	Use the pdobils -U -C -H command. Execute this command and use a redirect (>) for output to the target file.	6	4	--
11	Server communication control information	Use the pdls -d rpc -a command. Execute this command at 60-second intervals and output to the target file using an appending redirect (>>). Save this file on a medium such as a DAT. Recommended collection interval: 60 seconds Disk space needed to collect information one time: approximately 62 KB.	5	4	4
12	Information on the status of server transactions	Use the pdls -d tm -a command. Execute this command at 60-second intervals and output to the target file using an appending redirect (>>). Save this file on a medium such as a DAT. Recommended collection interval: 60 seconds Disk space needed to collect information one time: approximately 30 KB	5	6	5
13	Information on the status of server processes	Use the pdls -d prc -a command. Execute this command at 60-second intervals and output to the target file using an appending redirect (>>). Save this file on a medium such as a DAT. Recommended collection interval: 60 seconds Disk space needed to collect information one time: approximately 28 KB	5	6	5
14	Information on server lock	Use the pdls -d lck -a command. Execute this command at 60-second intervals and output to the target file using an appending redirect (>>). Save this file on a medium such as a DAT. Recommended collection interval: 60 seconds Disk space needed to collect information one time: approximately 62 KB	5	--	--
15	Information on the global buffer usage status	Use the pdbufls command. Execute this command at 60-second intervals and output to the target file using an appending redirect (>>). Save this file on a medium such as a DAT. Recommended collection interval: 60 seconds Disk space needed to collect information one time: approximately 3 KB	5	--	--
16	SQL trace files and error log files	SQL trace files and error log files are output to the directory specified by the PDCLTPATH operand of the client environment definition. When the PDCLTPATH operand is omitted, these files are output to the directory in which the AP is currently running. Save the output files on a medium such as a DAT. File names start with pderr or pdsql.	--	6	5
17	System log files	Use pdlogunld to unload the system log. Save the unload log files on a medium such as a DAT.	6	7	6
18	Core files	If a utility stops responding, use the following procedure to collect the core files: Procedure Use the pdls -d prc command to check the server process ID of the utility. Use the pdcancel -d command to cancel the server process identified in step 1.	--	3	--

Legend:

Prf.: Performance problem

NR: No-response

AT: Abnormal termination

--: Collection of this information is not necessary.

#1: Use the following formula to determine the required disk capacity:

n: Number of SQL objects stored in the SQL object buffer

For the number of SQL objects stored in the SQL object buffer, use the sum of the average number of SQL objects in the SQL object buffer and the average number of objects for stored procedures in the statistical information related to system operations. For details about the statistical information related to system operations, see Statistics Analysis Utility (pdstedit) in the manual HiRDB Version 9 Command Reference.

#2: Use the following formula to determine the required disk capacity:

500 + SQL object size (bytes)

For the SQL object size, use the value of SIZE (size) displayed by the pdobils command with the -N option specified. For details about SIZE (size), see the description of the pdobils command in the manual HiRDB Version 9 Command Reference.

#3: Use the following formula to determine the required disk capacity:

n: Number of UAPs executing SQL statements

For the number of UAPs executing SQL statements, use the value of the pd_max_users operand.

Notes

• The necessary disk space is a reference value. The actual value will depend on the system configuration.
• Available disk space is suppressed because files that can be added to by redirecting output information increase in size incrementally. Therefore, create a general-purpose shell script that swaps files and reuses them in a specific generation.