IBM Support

The IBM PureData System for Analytics database restore fails with 'Broken pipe' and 'Operation aborted'

Technote (troubleshooting)


Problem(Abstract)

PureData System for Analytics database restore fails with 'ERROR: 0 : Error uncompressing data file: output buffer too small; probable cause being corrupted input data'

Symptom

When the nzrestore is inserting data into the database it returns an error and fails. Here are excerpts with details for each relevant log:

/nz/kit/log/postgres/pg.log
2014-06-11 10:13:42.512488 EDT QUERY: insert into TABLE1 select * from external '/mnt/backup/NZ1/Netezza/NZ1/DB1/20140426012345/1/FULL/data/12345678.full.3.1' using( format 'internal' compress true );
ERROR: 0 : Error uncompressing data file: output buffer too small; probable cause being corrupted input data


/nz/kit/log/restoresvr/restoresvr.<pid>.<date>.log
2014-06-11 10:41:24.928167 EDT Info: Restoring TABLE1
2014-06-11 10:41:24.662253 EDT Info: severity=1 errno=32 /nzscratch/tmp/bnr10254DATrbq: Broken pipe in /production/builds/rel-6.0.5-P17/111223-20492rel-6.0.5-P17/main/src/bnr/nbclib/nbcConnectorHelper.cpp at line 1508
2014-06-11 10:41:24.662582 EDT Error: NZ-01844: /nzscratch/tmp/bnr10254DATrbq Broken pipe.
2014-06-11 10:41:24.667343 EDT Error: NZ-01114/01115: DB SQL 'insert into TABLE1 select * from external '/nzscratch/tmp/bnr10254DATrbq' using( format 'internal' compress true );' failed - DB error - ERROR: 0 : Error uncompressing data file: output buffer too small; probable cause being corrupted input data.
2014-06-11 10:41:24.667649 EDT Info: Operation aborted
2014-06-11 10:41:24.671559 EDT Info: NZ-00023: --- program 'restoresvr' (10254) exiting on host 'nznftest2tf18' ... ---


/nz/kit/log/dbos/dbos.log
2014-06-11 10:41:25.578007 EDT Info: plan queued: planid 6789 tx 0x23a3b cli 367 uid 2345 sid 54321 pid [12345]
2014-06-11 10:41:25.578198 EDT Info: plan prep : planid 6789 tx 0x23a3b cli 367 uid 2345 sid 54321 pid [12345]
2014-06-11 10:41:25.583749 EDT Info: plan in GRA: planid 6789 tx 0x23a3b cli 367 uid 2345 sid 54321 pid [12345]
2014-06-11 10:41:25.583795 EDT Info: plan start : planid 6789 tx 0x23a3b cli 367 uid 2345 sid 54321 pid [12345] job new
2014-06-11 10:41:25.614752 EDT Warning: Reload dataslice count mismatch, uncompress on host.
2014-06-11 10:42:29.791879 EDT Error: Caught unexpected CError in handleLoadCommandDataReader tx=0x23a3b
: *** (dbosEvent) 0: Error uncompressing data file: output buffer too small; probable cause being corrupted input data in /production/builds/rel-6.0.5-P17/111223-20492rel-6.0.5-P17/main/src/nde/misc/nzzlib.cpp at line 202
2014-06-11 10:42:29.791942 EDT Debug: NzSendErrorToClient: error code N-74 sent to cli 367 uid 2345 sid 54321 pid [12345]
2014-06-11 10:42:29.792272 EDT Info: plan abort : planid 6789 tx 0x23a3b cli 367 uid 2345 sid 54321 pid [12345] spu ID 0 err = 0 : Error uncompressing data file: output buffer too small; probable cause being corrupted input data
2014-06-11 10:42:29.792679 EDT Debug: tx 0x23a3b cli 367 uid 2345 sid 54321 pid [12345] setState from Active (0x1) to StartAbort (0x8)


Cause

This issue is still under investigation about how a datafile can became corrupt.

Environment

Any IBM PureData System for Analytics with any NPS

Resolving the problem

Possible ways to resolve this situation:


1) Restore everything except for the table where the restore fails. Then contact IBM Support to look for options for the missing tables.

2) Take a new full backup and restore that to the target. This is the preferred solution.

Product Alias/Synonym

NPS Netezza Platform Software

Document information

More support for: PureData System for Analytics
Database

Software version: 1.0.0

Operating system(s): Platform Independent

Reference #: 1671540

Modified date: 13 August 2014


Translate this page: