I won't convert thread style, so, i'm top-posting :)
All this would be great if it worked. The main problem is that
temporary directory option for mysqld iS NOT USED while doing
table repair. All files while repairing tables are placed
in the same directoty with the same owner and group as the original
database files (i mean, for example, .TMD files which can be huge!).
So, i am currenctly stuck with no solultion to the problem.
--
Artem
Russell E Glaue wrote:
Using Quota on Unix, you can specify a different quota soft_limit for
temporary space.
You can set MySQL to use /tmp as the temporary space, then set a
soft_limit of 10mb and a hard_limit of 300MB.
This quota should be separate from the normal user space quota.
With MySQL temp dir set to /tmp, and a soft/hard quota as 10MB/300MB,
the recovery should occur successfully.
And going above the soft_limit gives the user X amount of days to get
back down below the soft_limit. And the user cannot go above the
hard_limit.
You can set the soft_limit recovery days to like 1 day. They can go
over the 10MB soft_limit for 1 day, up to the hard_limit of 300MB,
but after one day the temp files will be deleted until the soft_limit
is satisfied.
This is a great way to control temp space, and not let users abuse
it, but still give them a lot of space to perform things like large
mysql database recoveries.
-RG
Artem Kuchin wrote:
I am asking this question as a hosting PROVIDER, not
as a hosting client.
The sitation is like this (this is a simple example with small
numbers):
1) a client has 100mb quota for site and db files (set on unix group)
2) 80mb is already taken by 75MB database in ONE TABLE and 5mb of
files 3) databases gets broken
4) mysql tried to auto repair the table
5) to repair data it must create a temporary file and this files take
almost as much as 75MB
6) user goes over quota becauase mysql needs another 75MB to reapir
table, and this summs up like 75+75+5 a lot > than 100mb in quota
7) mysql gets stuck on this db
8) people continue to acccess site and connections build up and each
connection waits for repair to be finished
9) eventually mysql runs out of connections and this stop other
clients too.
10) day is ruined.
SO the questions are:
1) Is the way to specify USER/GROUP and PLACE for temporary files
for table repair?
2) Maybe there is some OTHER way to set quota for databases?
3) Is there way to specify max number of connections for a user?
Current solution for this as i see it is to make a daemon which
must connect to database first with administrative privs and
monitor number of connections and repairs using processlist. If some
user does more than allowed number of connection then just kill them
off. If repair accures - then turn off that host for a while and
turn if back on again
when repair finished.
I wonder how it is solved on mass hosting servers.
--
Regards,
Artem
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
