Bug #11130 crash of datanodes on disk full

• Submitted: 7 Jun 2005 9:17
• Modified: 8 Jun 2005 12:32
• Reporter: Jan Kneschke
• Status: Duplicate
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S3 (Non-critical)
• Version: 4.1.12
• OS: Linux (Linux/EM64T (RHEL4))
• CPU Architecture: Any

[7 Jun 2005 9:17] Jan Kneschke

Description:
If the disk is full the datanodes are crashing when you try to import new data.

Date/Time: Tuesday 7 June 2005 - 10:26:16
Type of error: error
Message: Internal program error (failed ndbrequire)
Fault ID: 2341
Problem data: DbtupFixAlloc.cpp
Object of reference: DBTUP (Line: 288) 0x0000000a
ProgramName: /usr/local/mysql-max-4.1.12-unknown-linux-gnu-x86_64-glibc23/bin/ndbd
ProcessID: 16670
TraceFile: /usr/local/mysql/data/ndb_4_trace.log.3
Version 4.1.12
***EOM***

How to repeat:
Fill up the disk where ndbd is storing its logs and create new tables/insert new data (and wait for the GCP).

Suggested fix:
Don't crash.

[8 Jun 2005 5:25] Jorge del Conde

Verified w/Linux FC2 ext3, 4.1 bk.

[8 Jun 2005 12:32] Jonathan Miller

Same as MySQL Bugs: #10709: Cluster crashed if disks are out of space