List:Cluster« Previous MessageNext Message »
From:Jonas Oreland Date:February 23 2005 7:24am
Subject:Re: NDB Cluster Problems
View as plain text  
FYI: I just replied to the bug report

/Jonas

Faisal Feroz wrote:
> i dont think that this is a memory related problem as i said earlier
> that i got the same error when i tried with NoOfReplicas set to 1 and
> other things remaining the same, also i filed a bug if ne one knows or
> had a similar problem plz help as im running out of time.
> 
> 
> On Tue, 22 Feb 2005 21:06:02 +0500, Faisal Feroz <faisalferoz@stripped>
> wrote:
> 
>>hi,
>>
>>I have been trying to get the dump imported on the ndb cluster since
>>last two weeks but every time something strange happens after which my
>>cluster fails to start and the only remedy is to start the storage
>>nodes with the --initial parameter which drops all of my tables.
>>
>>im willing to do anything to get the cluster working ...
>>
>>The size of the dump is around 4GB and during the dump following error
>>occurs after which everything goes down, by the size of the dump i
>>mean the size of the dump file created by mysqldump.
>>
>>ERROR 1297 (HY000): Got Temporary error 4010 'Node failure caused
>>abort of transaction' from ndb cluster
>>
>>after the above error every thing just goes down and the following
>>errors just keep on coming
>>ERROR 1015 (HY000): Can't lock file (errorno: 4009) which i think is
>>related to the storage nodes going down.
>>
>>the architecture of the cluster is that we have two servers each
>>running the mysqld and ndbd process and one server acting as the
>>manager ...
>>before that we also tried setting the NoOfReplicas=1 and started
>>importing the dump  but the same errors followed after which we tried
>>to do the whole thing again with setting NoOfReplicas=2 but no
>>progress.
>>
>>The error occurs when last 20-30 tables are left we have a total of
>>186 tables. I dont think that this is related to the memory coz as i
>>said earlier i tried the same thing with only changing the
>>NoOfReplicas to 1
>>
>>the configurations are as follows:
>>mysql version: mysql-max-4.1.10-sun-solaris2.9-sparc-64bit
>>
>>config.ini follows
>>
>>[NDBD DEFAULT]
>>NoOfReplicas=2
>>IndexMemory=756M
>>DataMemory=3072M
>>MaxNoOfAttributes=10000
>>MaxNoOfOrderedIndexes=8000
>>MaxNoOfUniqueHashIndexes=5000
>>[MYSQLD DEFAULT]
>>[NDB_MGMD DEFAULT]
>>[TCP DEFAULT]
>># Managment Server
>>[NDB_MGMD]
>>HostName=192.168.10.35    # the IP of THIS SERVER
>># Storage Engines
>>[NDBD]
>>HostName=192.168.10.1 # the IP of the FIRST SERVER
>>DataDir= /var/lib/mysql-cluster
>>[NDBD]
>>HostName=192.168.10.2           # the IP of the SECOND SERVER
>>DataDir=/var/lib/mysql-cluster
>># 2 MySQL Clients
>>[MYSQLD]
>>[MYSQLD]
>>
>>the error log files are as follows:
>>[for storage node 2 -- 192.168.10.2]
>>
>>Date/Time: Tuesday 22 February 2005 - 18:23:48
>>Type of error: error
>>Message: Internal program error (failed ndbrequire)
>>Fault ID: 2341
>>Problem data: DbtupGen.cpp
>>Object of reference: DBTUP (Line: 484) 0x0000000e
>>ProgramName: ndbd
>>ProcessID: 13743
>>TraceFile: /var/lib/mysql-cluster/ndb_3_trace.log.10
>>***EOM***
>>
>>[for storage node  1 -- 192.168.10.1]
>>Date/Time: Tuesday 22 February 2005 - 12:47:53
>>Type of error: error
>>Message: File has already been opened
>>Fault ID: 2807
>>Problem data:
>>Object of reference: OpenFiles::insert()
>>ProgramName: ndbd
>>ProcessID: 13639
>>TraceFile: /var/lib/mysql-cluster/ndb_2_trace.log.7
>>***EOM***
>>
>>Date/Time: Tuesday 22 February 2005 - 12:51:31
>>Type of error: error
>>Message: No message slogan found
>>Fault ID: 0
>>Problem data:
>>Object of reference: DBDICT (Line: 0) 0x0000000a
>>ProgramName: ndbd
>>ProcessID: 13709
>>TraceFile: /var/lib/mysql-cluster/ndb_2_trace.log.8
>>***EOM***
>>
>>The specs fo the servers running the storage nodes are as follows:
>>192.168.10.1
>>Ram: 8GB
>>HD: SCSI 80GB
>>Sun SPARC Enterprise 3500
>>OS: Solaris 10
>>running ndbd, mysqld
>>
>>192.168.10.2
>>Ram: 8GB
>>HD: SCSI 80GB
>>Sun SPARC Enterprise 3500
>>OS: Solaris 9
>>running ndbd, mysqld
>>
>>192.168.10.35
>>Ram: 2GB
>>HD: SCSI 40GB
>>Sun SPARC Enterprise 420R
>>OS: Solaris 10
>>running ndb_mgmd
>>
> 
> 


-- 
Jonas Oreland, Software Engineer
MySQL AB, www.mysql.com
Thread
NDB Cluster ProblemsFaisal Feroz22 Feb
  • Re: NDB Cluster ProblemsAlex Davies22 Feb
  • Re: NDB Cluster ProblemsMikael Ronström22 Feb
  • Re: NDB Cluster Problemspekka22 Feb
  • Re: NDB Cluster ProblemsFaisal Feroz23 Feb
    • Re: NDB Cluster ProblemsJonas Oreland23 Feb
  • Re: NDB Cluster ProblemsFaisal Feroz23 Feb