Hi,
Mark Malakanov wrote:
> I see. Thank you Adam.
>
> The reason I asked I had problems to load a table
> with 1 mln rows by "ALTER TABLE" and by "INSERT ..
> SELECT" methods.
> Always got errors like "REDO is not enough.
This is a typical error when performing high write load.
You need to:
1) Increase NoOfFragmentLogFiles
2) Increase NoOfDiskPagesToDiskAfterRestartTUP/ACC
To calculate amounts check one of my blog entries at
mikaelronstrom.blogspot.com
You might require fairly substantial numbers for number
of fragment log files given that the load on the log at
inserts can be fairly high.
INSERT ... SELECT will most likely suffer from transaction failure
since it is attempted as one transaction.
Rgrds Mikael
> Decrease TimeBetweenLocalCheckpoints or|and
> increase NoOfFragmentLogFiles."
> and even "unknown error".
> I decreased TimeBetweenLocalCheckpoints, increased
> NoOfFragmentLogFiles - but no luck.
>
> My config.ini just in case.
>
>
> NoOfReplicas=1
> DataMemory=2000M
> IndexMemory=200M
> MaxNoOfConcurrentOperations=200000
> NoOfFragmentLogFiles=16
> # 18 - 1Mb; 19 - 2Mb; 20 - 4Mb ...
> TimeBetweenLocalCheckpoints=17
> #TimeBetweenGlobalCheckpoints = 1500
> RedoBuffer=256M
> #NoOfDiskPagesToDiskAfterRestartTUP = 175
> #NoOfDiskPagesToDiskAfterRestartACC = 80
>
>
>
>
>
> # Managment Server
>
> # the IP of THIS SERVER
> HostName=192.168.0.21
>
> # Storage Engines
>
>
>
> I have vorked it around by a procedure with "loop
> fetch; insert; end loop;"
> It worked fine and quite fast.
Mikael Ronstrom
Senior Software Architect, MySQL AB
My blog:
http://mikaelronstrom.blogspot.com