List Info

Thread: Massive Performance Problems




Massive Performance Problems
user name
2006-05-26 16:49:54
Hi,
Try setting
ndb_use_exact_count=0
Should help short queries.
The delete * from .... that take 20 seconds need more
information in 
order
to understand whether it is reasonable or not.
Do I understand it correctly that data nodes are not sharing
servers 
with others?
Have you measured CPU load on the various servers involved
during e.g.
the Delete query.

Rgrds Mikael

2006-05-26 kl. 13.50 skrev Andre Podschaske:

> Hi,
>
> I have recently installed a Mysql Cluster with 2
Storage Nodes, 2 Mysql
> Daemons and some ndb Manager.
> Now i encounter severe performance lags.
>
> The following hardware is all alike in the machines
used:
>
> Dual/Dual Opteron 270
> 8GB ECC Ram
> 8x80 GB SATA Raid1 HDD
> 2x1000MBit Ethernet
>
> Mysql Version 5.0.21 max
>
> The Clusterpart of the My.cfg of the Mysqld´s:
>
> ndbcluster
> ndb-connectstring=192.168.2.90
> engine_condition_pushdown=1
>
> The ndb Config.ini:
>
> [ndbd default]
> NoOfReplicas= 2
> MaxNoOfConcurrentOperations= 500000
> MaxNoOfLocalOperations= 550000
> DataMemory= 3500M
> IndexMemory= 1000M
> TimeBetweenWatchDogCheck= 20000
> DataDir= /var/lib/mysql-cluster
> MaxNoOfOrderedIndexes= 6000
> MaxNoOfUniqueHashIndexes= 6000
> MaxNoOfTables= 700
> MaxNoOfAttributes= 3000
> TimeBetweenLocalCheckpoints= 20
> TransactionDeadlockDetectionTimeout=6000
> TransactionBufferMemory=3M
> MaxNoOfConcurrentScans=300
> MaxNoOfLocalScans=600
> RedoBuffer=128M
> UndoDataBuffer=128M
> UndoIndexBuffer=8M
> StartPartialTimeout= 30000
> StartPartitionedTimeout= 60000
> TimeBetweenGlobalCheckpoints=1500
> TimeBetweenInactiveTransactionAbortCheck=5000
>
> [ndb_mgmd default]
> DataDir= /var/lib/mysql-cluster
>
> [ndb_mgmd]
> HostName= 192.168.2.90
>
> [ndbd]
> HostName= 192.168.2.19
>
> [ndbd]
> HostName= 192.168.2.3
>
> [mysqld]
> HostName= 192.168.2.4
> [mysqld]
> HostName= 192.168.2.5
>
> I am already trying for days, but i dont get a useable
performance 
> output.
> A simple delete * from data where uid = 'test' in a
table takes up to 
> 20
> seconds.
> Additionally, as more people access the cluster it gets
slower and 
> slower
> until browser-timeout.
> As the database is not that large in size(600MB) i
expected to reach 
> min.
> 80% of the performance of a single-mysql.
> I have changed and tried lots of things in the config,
but i didnt get
> positive results.
> Does anyone of you has an idea on how to improve the
performance?
>
> kind regards
>
> Andre Podschaske
>
>
>
>
> -- 
> MySQL Cluster Mailing List
> For list archives: http://lists.mysql.com
/cluster
> To unsubscribe:    
> http://li
sts.mysql.com/cluster?unsub=mikaelmysql.com
>
>
Mikael Ronstrom, Senior Software Architect
MySQL AB, www.mysql.com

Jumpstart your cluster:
http://www.mysql.com/consulting/packaged/cluster.html
My blog:
http://mikaelronst
rom.blogspot.com


-- 
MySQL Cluster Mailing List
For list archives: http://lists.mysql.com
/cluster
To unsubscribe:    http://list
s.mysql.com/cluster?unsub=bondyahoo.com

Massive Performance Problems
user name
2006-05-30 18:10:00
You could use truncate table instead of delete from.

Jonathan Miller
Austin, Texas USA
Senior Quality Assurance Developer
MySQL AB www.mysql.com
    __  ___     ___ ____  __
   /  |/  /_ __/ __/ __ \/ /  
  / /|_/ / // /\ \/ /_/ / /__ 
 /_/  /_/\_, /___/\___\_\___/ 
        <___/   www.mysql.com 

Jumpstart your cluster!
http://www.mysql.com/consulting/packaged/cluster.html

Get training on clusters
http://www.mysql.com/training/courses/mysql_cluster.html


All-in-one Enterprise-grade Database, Support and Services
http://www.mysql.com/ne
twork/

-> -----Original Message-----
-> From: Mikael Ronström [mailto:mikaelmysql.com]
-> Sent: Friday, May 26, 2006 11:50 AM
-> To: Andre Podschaske
-> Cc: clusterlists.mysql.com
-> Subject: Re: Massive Performance Problems
-> 
-> Hi,
-> Try setting
-> ndb_use_exact_count=0
-> Should help short queries.
-> The delete * from .... that take 20 seconds need more
information in
-> order
-> to understand whether it is reasonable or not.
-> Do I understand it correctly that data nodes are not
sharing servers
-> with others?
-> Have you measured CPU load on the various servers
involved during e.g.
-> the Delete query.
-> 
-> Rgrds Mikael
-> 
-> 2006-05-26 kl. 13.50 skrev Andre Podschaske:
-> 
-> > Hi,
-> >
-> > I have recently installed a Mysql Cluster with 2
Storage Nodes, 2 Mysql
-> > Daemons and some ndb Manager.
-> > Now i encounter severe performance lags.
-> >
-> > The following hardware is all alike in the
machines used:
-> >
-> > Dual/Dual Opteron 270
-> > 8GB ECC Ram
-> > 8x80 GB SATA Raid1 HDD
-> > 2x1000MBit Ethernet
-> >
-> > Mysql Version 5.0.21 max
-> >
-> > The Clusterpart of the My.cfg of the Mysqld´s:
-> >
-> > ndbcluster
-> > ndb-connectstring=192.168.2.90
-> > engine_condition_pushdown=1
-> >
-> > The ndb Config.ini:
-> >
-> > [ndbd default]
-> > NoOfReplicas= 2
-> > MaxNoOfConcurrentOperations= 500000
-> > MaxNoOfLocalOperations= 550000
-> > DataMemory= 3500M
-> > IndexMemory= 1000M
-> > TimeBetweenWatchDogCheck= 20000
-> > DataDir= /var/lib/mysql-cluster
-> > MaxNoOfOrderedIndexes= 6000
-> > MaxNoOfUniqueHashIndexes= 6000
-> > MaxNoOfTables= 700
-> > MaxNoOfAttributes= 3000
-> > TimeBetweenLocalCheckpoints= 20
-> > TransactionDeadlockDetectionTimeout=6000
-> > TransactionBufferMemory=3M
-> > MaxNoOfConcurrentScans=300
-> > MaxNoOfLocalScans=600
-> > RedoBuffer=128M
-> > UndoDataBuffer=128M
-> > UndoIndexBuffer=8M
-> > StartPartialTimeout= 30000
-> > StartPartitionedTimeout= 60000
-> > TimeBetweenGlobalCheckpoints=1500
-> > TimeBetweenInactiveTransactionAbortCheck=5000
-> >
-> > [ndb_mgmd default]
-> > DataDir= /var/lib/mysql-cluster
-> >
-> > [ndb_mgmd]
-> > HostName= 192.168.2.90
-> >
-> > [ndbd]
-> > HostName= 192.168.2.19
-> >
-> > [ndbd]
-> > HostName= 192.168.2.3
-> >
-> > [mysqld]
-> > HostName= 192.168.2.4
-> > [mysqld]
-> > HostName= 192.168.2.5
-> >
-> > I am already trying for days, but i dont get a
useable performance
-> > output.
-> > A simple delete * from data where uid = 'test'
in a table takes up to
-> > 20
-> > seconds.
-> > Additionally, as more people access the cluster
it gets slower and
-> > slower
-> > until browser-timeout.
-> > As the database is not that large in size(600MB)
i expected to reach
-> > min.
-> > 80% of the performance of a single-mysql.
-> > I have changed and tried lots of things in the
config, but i didnt get
-> > positive results.
-> > Does anyone of you has an idea on how to improve
the performance?
-> >
-> > kind regards
-> >
-> > Andre Podschaske
-> >
-> >
-> >
-> >
-> > --
-> > MySQL Cluster Mailing List
-> > For list archives: http://lists.mysql.com
/cluster
-> > To unsubscribe:
-> > http://li
sts.mysql.com/cluster?unsub=mikaelmysql.com
-> >
-> >
-> Mikael Ronstrom, Senior Software Architect
-> MySQL AB, www.mysql.com
-> 
-> Jumpstart your cluster:
-> http://www.mysql.com/consulting/packaged/cluster.html
-> My blog:
-> http://mikaelronst
rom.blogspot.com
-> 
-> 
-> --
-> MySQL Cluster Mailing List
-> For list archives: http://lists.mysql.com
/cluster
-> To unsubscribe:
-> http://
lists.mysql.com/cluster?unsub=jbmillermysql.com


-- 
MySQL Cluster Mailing List
For list archives: http://lists.mysql.com
/cluster
To unsubscribe:    http://list
s.mysql.com/cluster?unsub=bondyahoo.com

Massive Performance Problems
user name
2006-06-01 01:10:33
On 31/05/2006 6:10 a.m., Jonathan Miller wrote:
> You could use truncate table instead of delete from.
> 

But be careful, I have often found that TRUNCATE TABLE on
large 
ndbcluster tables causes cluster failure...

-Simon

-- 
MySQL Cluster Mailing List
For list archives: http://lists.mysql.com
/cluster
To unsubscribe:    http://list
s.mysql.com/cluster?unsub=bondyahoo.com

Massive Performance Problems
user name
2006-06-01 03:31:46
If you get a failure truncating tables, please open a bug
and let me know
about it.

Thanks.


Jonathan Miller
Austin, Texas USA
Senior Quality Assurance Developer
MySQL AB www.mysql.com
    __  ___     ___ ____  __
   /  |/  /_ __/ __/ __ \/ /  
  / /|_/ / // /\ \/ /_/ / /__ 
 /_/  /_/\_, /___/\___\_\___/ 
        <___/   www.mysql.com 

Jumpstart your cluster!
http://www.mysql.com/consulting/packaged/cluster.html

Get training on clusters
http://www.mysql.com/training/courses/mysql_cluster.html


All-in-one Enterprise-grade Database, Support and Services
http://www.mysql.com/ne
twork/

-> -----Original Message-----
-> From: Simon Garner [mailto:sgarnerexpio.co.nz]
-> Sent: Wednesday, May 31, 2006 8:11 PM
-> To: clusterlists.mysql.com
-> Subject: Re: Massive Performance Problems
-> 
-> On 31/05/2006 6:10 a.m., Jonathan Miller wrote:
-> > You could use truncate table instead of delete
from.
-> >
-> 
-> But be careful, I have often found that TRUNCATE TABLE
on large
-> ndbcluster tables causes cluster failure...
-> 
-> -Simon
-> 
-> --
-> MySQL Cluster Mailing List
-> For list archives: http://lists.mysql.com
/cluster
-> To unsubscribe:
-> http://
lists.mysql.com/cluster?unsub=jbmillermysql.com


-- 
MySQL Cluster Mailing List
For list archives: http://lists.mysql.com
/cluster
To unsubscribe:    http://list
s.mysql.com/cluster?unsub=bondyahoo.com

[1-4]

about | contact  Other archives ( Real Estate discussion Medical topics )