[Info-vax] Databases versus RMS

[JF Mezei]

Arne Vajhøj wrote:

> With InnoDB tables you only have a few files for the entire
> database no matter how many tables.

This is what I get with only 1 database and 3 tables.  those
mysql-bin.xxxxxxx files seems to contain the data. the "bike" database
conftains what seems to be just the table definitions (.frm files)




> velo:mysql $  ls
> bike                    mysql-bin.000005        mysql-bin.000014        mysql-bin.000023        mysql-bin.index
> ib_logfile0             mysql-bin.000006        mysql-bin.000015        mysql-bin.000024        mysql.sock
> ib_logfile1             mysql-bin.000007        mysql-bin.000016        mysql-bin.000025        mysql_service.log
> ibdata1                 mysql-bin.000008        mysql-bin.000017        mysql-bin.000026        mysql_upgrade_info
> mysql                   mysql-bin.000009        mysql-bin.000018        mysql-bin.000027        performance_schema
> mysql-bin.000001        mysql-bin.000010        mysql-bin.000019        mysql-bin.000028        velo.vaxination.ca.pid
> mysql-bin.000002        mysql-bin.000011        mysql-bin.000020        mysql-bin.000029
> mysql-bin.000003        mysql-bin.000012        mysql-bin.000021        mysql-bin.000030
> mysql-bin.000004        mysql-bin.000013        mysql-bin.000022        mysql-bin.000031
> 
> velo:mysql $  


Until  I understand how it uses those files and how the data is
organised between the multiple files, I can't claim to understand enough
of this database system to support a production one.

BTW, that dslreports.domc web site is still down.

When LDDRIVER fucked up a bound volume set for me, It took me a while,
but I was able to recover much (but not all) because the ODS file system
is well documented and I had help from folks like Hein and Hoff and I
was even able to recover lost records from an indexed file (parts of
index was zapped).

In he xase of DSLREPORTS, it appears the guy can't recover even though
he confirms the data appears to still be all there. Not sure if due to
lack fo experience or just because the combination of storage arrays and
DBMS systems makes things just impossible to fix when a proble occurs.