Bug #29177 crash on disk data >1GB and node restart
Submitted: 18 Jun 2007 14:29 Modified: 21 Jun 2007 9:23
Reporter: Pekka Nousiainen Email Updates:
Status: Duplicate Impact on me:
None 
Category:MySQL Cluster: Disk Data Severity:S2 (Serious)
Version:5.1 OS:Linux (64-bit)
Assigned to: CPU Architecture:Any

[18 Jun 2007 14:29] Pekka Nousiainen 
Description:
various crashes

1)
Error data: dbtup/DbtupDiskAlloc.cpp
Error object: DBTUP (Line: 549) 0x0000000a

context:
    pageBits= tsman.alloc_page_from_extent(&ext.p->m_key, bits);
    jamEntry();
    ddassert(pageBits >= 0);

2)
Error data: Invalid memory access: ptr (11801a 0x1020078) magic: (ffffff00 00000
082) memroot: 0xbc0000 page: 41
Error object: DBTUP (Line: 41) 0x0000000a

context: line 636
disk_page_prealloc_dirty_page
...
634       ddassert(free >= used);
635       Ptr<Extent_info> extentPtr;
636       c_extent_pool.getPtr(extentPtr, ext); //<-- here, bad magic

How to repeat:
2 nodes
create tablespace with one datafile > 1GB
hugoLoad a table to > 90% full
restart slave node (not initial)
drop table
hugoLoad..
crash

Suggested fix:
extent info corruption at NR ?
[21 Jun 2007 9:23] Jonas Oreland 
duplicate of bug#29229 (or other way around)