Bug #6871 DBD execute failed: Got error 897 'Unknown error code' from ndbcluster
Submitted: 29 Nov 2004 10:10 Modified: 1 Dec 2004 9:26
Reporter: Martin Boer Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S1 (Critical)
Version:4.1.7 OS:Linux (Debian (stable, partly testing))
Assigned to: Jonas Oreland CPU Architecture:Any

[29 Nov 2004 10:10] Martin Boer
Ndbd crashes with an error 897 when running my script. 

I have a setup with 2 linuxmachines (compaq dl360, 2x3.0GHz, 2G ram, 15k scsi discs each). 
The config.ini looks like this: 

NoOfReplicas= 2
DataDir= /afs2/mysql/4.1.7/mysql-cluster
IndexMemory = 400MB
DataMemory = 1000MB
MaxNoOfConcurrentOperations = 100000
MaxNoOfOrderedIndexes = 50000
MaxNoOfUniqueHashIndexes = 256
Hostname= pwbeat06.beat.bnet.nl
HostName= pwbeat06.beat.bnet.nl
HostName= hdbeat05.beat.bnet.nl
HostName= pwbeat06.beat.bnet.nl
HostName= hdbeat05.beat.bnet.nl

The database has 44 tables, but is quite small (now 16Mb, will grow up to 80 when it stays in MyISAM). The thing is that over 80% of the data will be updated every 10 minutes. 

When I convert the tables to ndb I can see all the tables on both API's. When I insert something on machine 1, and select it on 2 it is right there. So it seems the cluster works

When I start (one of) my script(s)  though, I get error 897's directly. After about 20 of those error's ndbd crashes (most of the times on both machines).

How to repeat:
I can give you one of my scripts and some tables if you want, the crash occurs every time I do it. I'm not sure how else to repeat it. Just mail me. 

Suggested fix:
Stop making ndbd crashing ? I dunno.
[1 Dec 2004 9:26] Jonas Oreland
The bug was replace with charsets and could be reproduced by the following
>create table t1(a varchar(10) primary key);
>insert into t1 values("jonas % ");
>replace into t1 values("jonas % ");
which gave 897 and an inconsistent index

Fixed in 4.1-bk