How To
Summary
How do I use sdcp to collect IBM MQ diagnostic data on HPE Integrity NonStop?
Objective
Environment
Steps
- Starting sdcp
- Collecting data for a case
- Choosing sections to gather
- Choosing queue managers
- Handling large files
- Sending data to IBM
- Cleaning up files
Starting sdcp
The sdcp command is an Open System Services (OSS) shell script for collecting IBM MQ diagnostic data. Before using runmqras you may wish to clean up MQ files to reduce the amount of data collected. Ensure the MQ bin directory is in your PATH or type the full path to sdcp to run it. The sdcp commmand should also have access to the /tmp directory or it will not be able to gather vproc information.
Checking access to /tmp on HPE NonStop
sh> ls ‑ld /tmp
drwxrwxrwt 1 SUPER.SUPER SUPER /tmp
sh> PATH="$PATH:/path/to/mqm/bin"
sh> sdcp ‑?
By default, sdcp stores its output file in the current working directory, so you should be outside the MQ OSS installation path opt/mqm and var/mqm directory trees, and outside any subdirectory /G/ when starting sdcp so that later sdcp runs won't collect old sdcp archives. Alternatively, you can use the ‑w option to specify a different working directory as described in the section on handling large files.
Collecting data for a case
IBM MQ has joined the new Support Community which uses cases rather than problem management records to track issues. Case numbers like TS123456789 have replaced PMR numbers like 12345,678,999.
The sdcp command does not yet recognize case numbers, which are slightly shorter than PMR numbers. You can work around this by passing your case number to sdcp and padding it with two extra characters. For example:
Setting the case number on HPE Integrity NonStop
sh> sdcp ‑p TS123456789zz
Alternatively, you can avoid the ‑p option entirely. The sdcp command will create a file like WMQSDCP‑date‑time.archive.tar.Z and print the file name when it finishes. Rename the file so it starts with your case number before sending the file to IBM. For example:
Setting the case number manually on HPE Integrity NonStop
sh> mv WMQSDCP‑190628‑083107.archive.tar.Z TS123456789‑sdcp‑190628‑083107.archive.tar.Z
Choosing sections to gather
The sdcp command does not have sections like the runmqras command does. However, there are three options for getting additional information. These options can be used together.
- You can use the ‑f option to collect full data from certain commands such as WHOHAS. This option will make sdcp run significantly slower and is not usually required.
- If you are running IBM MQ V8.1, you can use the ‑x option to make sdcp collect diagnostic FDC files created by the amqxdbg program. This option may be particularly useful in cases where IBM MQ programs are unresponsive or running slowly.
- Finally, you can use the ‑d option to generate debug output from sdcp itself. This is required only when there is a problem with the sdcp command itself.
Collecting full WHOHAS data on HPE Integrity NonStop
sh> sdcp ‑p TS123456789zz ‑f
Collecting amqxdbg FDCs on HPE Integrity NonStop (MQ V8.1)
sh> sdcp ‑p TS123456789zz ‑x
Collecting sdcp debug output on HPE Integrity NonStop
sh> sdcp ‑p TS123456789zz ‑d
Choosing queue managers
By default the sdcp command tries to collect information about all queue managers in your MQ installation. Use the ‑m option to name a single queue manager sdcp should examine.
If you want to run sdcp against several queue managers, you must run it more than once with a different queue manager name each time.
The queue managers you choose should be running, or else some the commands issued by sdcp will fail with the error AMQ8146: IBM MQ queue manager not available. However, if a queue manager is running but unresponsive, the ‑e option tells sdcp not to run MQ commands against it so that sdcp itself will not hang.
Choosing queue managers on HPE Integrity NonStop
sh> sdcp ‑p TS123456789zz ‑m QMA
sh> sdcp ‑p TS123456789zz ‑m QMB
Handling large files
The tar archive created by sdcp can be very large if your system has lots of FDCs or trace files to collect. Normally sdcp uses space in the current working directory to collect and zip up the files.
To choose a different directory on a file system with more free space, use the ‑w option. The directory you give must exist and should be empty. The directory should be outside the MQ OSS installation opt/mqm and var/mqm directory trees so that later sdcp runs won't collect your sdcp archive file. For example:
Handling large files on HPE Integrity NonStop
sh> mkdir /var/bigdata/2019-07-27
sh> sdcp ‑p TS123456789zz ‑m QMA ‑w /var/bigdata/2019-07-27
Sending data to IBM
If sdcp cannot collect files due to security reasons, for example FFST files created by application programs, it will identify the files by name. Collect these files manually, as well as any ZZSAxxxx SAVEABEND files generated by MQ, in a zip or tar archive starting with the case number, for example TS123456789-extra-files.tar.Z.
Make sure the sdcp archive file starts with your IBM case number, for example TS123456789‑sdcp‑190628‑083107.archive.tar.Z. Then send the sdcp file and any extra files to IBM using any of the methods described here.
Blue Diamond clients must use the secure Blue Diamond upload instead.
Cleaning up files
The sdcp command does not delete any files from your system, neither MQ logs, nor FDCs, SAVEABEND files, or trace files. After collecting these files with sdcp, IBM recommends that you archive or delete your FDCs, SAVEABEND, and trace files. If you need to collect sdcp at a later time, the new sdcp file will be smaller and easier to analyze because it does not contain duplicate files and old information.
Keep a copy of your sdcp archive file after sending it to IBM until your problem is resolved and you have tested the solution to your satisfaction.
Related Information
Document Location
Worldwide
Was this topic helpful?
Document Information
Modified date:
08 April 2020
UID
ibm10957717