Am 25. Jul, 2010 schwätzte Alex Dean so:
On Jul 25, 2010, at 3:03 AM, der.hans wrote:
I'm needing to convert a very busy production myisam table that is
somewhat humongous to innodb and the conversion takes longer than the
maintenance windows.
If you have a slave which is capable of
Am 25. Jul, 2010 schwätzte Technomage so:
There is a program in OpenBSd that can help with this. its called carp. you
have 2 machines running, one the primary and the other the backup. if the
main fails, carp kicks in and take possession if the relevant ip addresses
(as far as your network is
On Jul 26, 2010, at 2:03 AM, der.hans wrote:
Am 25. Jul, 2010 schwätzte Technomage so:
There is a program in OpenBSd that can help with this. its called
carp. you have 2 machines running, one the primary and the other
the backup. if the main fails, carp kicks in and take possession if
Am 24. Jul, 2010 schwätzte Alex Dean so:
moin moin,
I think you should consider replicating to multiple slaves. 1 is for backup,
and is totally read-only.
That's my goal. First I need to get all of the DBen going to at least one
slave and make sure we've backing up everything that needs to
Am 24. Jul, 2010 schwätzte Bryan O'Neal so:
I agree with Alex but this just follows logical design of separation.
You dev, qa, prod, ha, and dr environments should be separated and
used for those set purposes. Unfortunately I also know that if this
most elementary step has not been take it is
Am 24. Jul, 2010 schwätzte Bryan O'Neal so:
Sadly Number of physical boxes is usually not the issue with resource
bottle neck; it is that you are always running close to capacity and
can not convince anyone to purchase more equipment because things are
running fine. Double the number of fsyncs
replication is not always perfect in MySQL - Innodb is less pron to
error but it still happens. Do your best to kill myisam.
What verifying myisam and innodb copies of the same data are exactly the
same? Actually, I want something that will diff them and apply the diff to
the innodb side.
Read only I believe is a server wide var, but again, I am not certain.
Replication should also be server wide , ie. no replication rules,
including which databases should be considered for replication. Just
recently learned what kind of problems it can cause otherwise.
On Sun, Jul 25, 2010 at
On Jul 25, 2010, at 3:03 AM, der.hans wrote:
I'm needing to convert a very busy production myisam table that is
somewhat humongous to innodb and the conversion takes longer than the
maintenance windows.
If you have a slave which is capable of becoming a production server,
you can convert
On 7/25/10 3:46 PM, Alex Dean wrote:
On Jul 25, 2010, at 3:03 AM, der.hans wrote:
I'm needing to convert a very busy production myisam table that is
somewhat humongous to innodb and the conversion takes longer than the
maintenance windows.
If you have a slave which is capable of becoming a
Egg.
Dinosaurs laid eggs long before there were chickens.
On Jul 23, 2010, at 9:59 PM, Bryan O'Neal wrote:
BTW - Their is an answer to the chicken and egg problem. Both - But
which answer your choose depends purely on if you are a creationist
or evolutionist at hart. I claim the scientists
On Jul 23, 2010, at 8:01 PM, der.hans wrote:
Am 23. Jul, 2010 schwätzte Alex Dean so:
Hm. Are users on the slave modifying tables which get replicated
from the master? That's grandma's recipe for data stew. You can't
do that if you
Yeah and yeah. I'm going to change that.
have any
I agree with Alex but this just follows logical design of separation.
You dev, qa, prod, ha, and dr environments should be separated and
used for those set purposes. Unfortunately I also know that if this
most elementary step has not been take it is usually due to a lack of
required resources not
On Jul 24, 2010, at 12:47 PM, Bryan O'Neal wrote:
I agree with Alex but this just follows logical design of separation.
You dev, qa, prod, ha, and dr environments should be separated and
used for those set purposes. Unfortunately I also know that if this
most elementary step has not been take
Sadly Number of physical boxes is usually not the issue with resource
bottle neck; it is that you are always running close to capacity and
can not convince anyone to purchase more equipment because things are
running fine. Double the number of fsyncs on a box and then you get a
complaint that
moin moin,
is there a way see when a particular grant was last used in MySQL? I can
probably troll logs, but I wonder if it's in the DB.
Also, how should I sync grants from master to slave when they're out of
sync?
In our case, the slave is also getting local changes. I'm pushing to stop
that,
On Jul 23, 2010, at 1:00 AM, der.hans wrote:
moin moin,
is there a way see when a particular grant was last used in MySQL? I
can
probably troll logs, but I wonder if it's in the DB.
I don't think that's logged anywhere.
Also, how should I sync grants from master to slave when they're
Other then query the MySQL grant table I am not sure of a good way to
reconcile recent changes. As for out of sync replication stop the
replication, apply a backup, start replicaiton.
Sorry I don't have better advice. I can try playing around with some
stuff this weekend and would love to know
With the general query log enabled in advance its easy to see when a
and by whom a query was run, just be sure to put that log in the
logrotate cycle because it can get big, fast..
Ben
On Fri, Jul 23, 2010 at 9:09 AM, Bryan O'Neal
bryan.on...@theonealandassociates.com wrote:
Other then query
I would not have that enabled on a production server, but you are correct
that if you do you can see almost everything
On Fri, Jul 23, 2010 at 12:03 PM, Ben Trussell azlob...@gmail.com wrote:
With the general query log enabled in advance its easy to see when a
and by whom a query was run, just
Come and live on the wild side Bryan!
On Fri, Jul 23, 2010 at 1:11 PM, Bryan O'Neal
bryan.on...@theonealandassociates.com wrote:
I would not have that enabled on a production server, but you are correct
that if you do you can see almost everything
On Fri, Jul 23, 2010 at 12:03 PM, Ben
Am 23. Jul, 2010 schwätzte Alex Dean so:
moin moin,
Also, how should I sync grants from master to slave when they're out of
sync?
Dump the 'mysql' database on the master, and load it on the slave.
If the slave has a different set of user accounts, I think you'll have to
manually create
Am 23. Jul, 2010 schwätzte Bryan O'Neal so:
Other then query the MySQL grant table I am not sure of a good way to
reconcile recent changes. As for out of sync replication stop the
replication, apply a backup, start replicaiton.
Time for an audit script.
Sorry I don't have better advice. I
Am 23. Jul, 2010 schwätzte Ben Trussell so:
With the general query log enabled in advance its easy to see when a
and by whom a query was run, just be sure to put that log in the
logrotate cycle because it can get big, fast..
Yeah, not activating that on prod :).
But, I might be able to block
I may also recommend that you don't do selective replication. Due to the
ability to manipulate data using fully qualified statements you can
circumvent replication rules. So it is best not to have them. I also highly
recommend you do not do circular replication, which is typically the next
Am 23. Jul, 2010 schwätzte Bryan O'Neal so:
I may also recommend that you don't do selective replication. Due to the
We're doing backups from slaves, so need full replication.
ability to manipulate data using fully qualified statements you can
circumvent replication rules. So it is best not
BTW - Their is an answer to the chicken and egg problem. Both - But which
answer your choose depends purely on if you are a creationist or
evolutionist at hart. I claim the scientists who decided the chicken came
first are creationists at hart even if they claim they are not. I say this
because
27 matches
Mail list logo