Manually Saving Teradata Crashdumps to Stream Files - Advanced SQL Engine - Teradata Database

Database Administration

Product
Advanced SQL Engine
Teradata Database
Release Number
17.05
17.00
Published
June 2020
Language
English (United States)
Last Update
2021-01-22
dita:mapPath
rgu1556127906220.ditamap
dita:ditavalPath
lze1555437562152.ditaval
dita:id
B035-1093
lifecycle
previous
Product Category
Teradata Vantageâ„¢

You can save crashdumps to flat (stream) files on each node involved in a crashdump. The stream files are in compressed binary FastLoad format. Stream files can either be transmitted directly to the Teradata Support Center or moved onto another system at the customer site, where debugging can occur. The cspstreamend program loads the dump data in the stream files into a table in the Crashdumps database for analysis.

DBAs can save crashdump files all at once or a few nodes at a time. Both SMP and MPP platforms can save crashdump stream files. Saving to stream files has the following benefits:
  • You do not need to save to or export from the Crashdumps database on the production system. This saves time.
  • Stream files enable you to move a dump to an on-premises test system for debugging. This frees the production system for other tasks.
  1. Decide if you want a crashdump to be saved to stream files by default or if you want to specify the save destination for each crashdump.
    If you want to make stream files the default, do the following:
    1. Create a custom resource file in /etc/opt/teradata/tdconfig/defaults/ named Csp.custom and add this line to the file: Target:stream.
    2. Issue a tpareset command.
      Subsequent csp -mode save commands automatically save to stream files. The customized resource file remains after an upgrade.
  2. When a crashdump occurs, save the raw dump files as stream files.
    For example, the following command tells CSP to save to stream files:
    > csp -mode save -target stream -force
          Dump 2017/05/12-05:22:30-06 is about to be saved
          Are you sure? [y,n,q,?] > y
  3. Collect the stream file from each node into a single directory.
    In the following example, the DBA collects the stream dump files named CRASH_XXX into the directory named /tmp/dump:
     pcl -collect /var/opt/teradata/tddump/pdedumps/stream/CRASH_20170512_052230_06 /tmp/dump
     All 3 node(s) have connected
            byn001-3:1023: collect completed: 234921696 bytes sent (1 files/0 directories)
            byn001-2:1023: collect completed: 257842344 bytes sent (1 files/0 directories)
            byn001-1:1022: collect completed: 308514708 bytes sent (1 files/0 directories)
  4. Decide where to send the crashdump stream files next:
    Choose one of the following:
    • The Teradata Support Center
    • A test or QA system
    If you want to load the stream files into the Crashdumps database on a test system, run the cspstreamend program to merge the stream files and load them onto the test system.
    In the following example, the DBA loads the data in the stream files from /tmp/dump into the Crashdumps database table named demodump on the test system named <Harold>.
    > cspstreamend -path /tmp/dump -dumpname demodump -system <Harold> 
    An alternative method of loading the stream files onto a test server is to copy the stream files to a directory on the test server. You can then run cspstreamend on the test server to load the dump to the Crashdumps database on the test server.