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 12:56 AM, der.hans <
PLUGd@lufthans.com> wrote:
> 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 be backed
> up.
>
>>
>> http://dev.mysql.com/doc/refman/5.1/en/server-system-variables.html#sysvar_read_only
>
> This is per DB? If db1 is a slave DB, I can set it as read-only, but keep
> db2, which is a local DB, as read-write?
>
>> Another slave, for reporting, can have more relaxed rules. If people want
>> to mess with the data there, no harm. If they screw it up too horribly,
>> restore a backup onto the reporting database and continue as before. You
>> need to get
>
> I still want the replicated parts to be read-only. Yes, I want reporting
> environments seperate from production slaves.
>
>> them to script any changes they're making, so they're easy to re-apply
>> after you restore the testing database. (If there are certain columns they
>> add, or aggregating tables they find useful, they should be able to run 1
>> script to re-create them. This removes the "we can't restore, it'll ruin
>> all my custom work!" complaint.)
>
> Great idea. This will be the hardest to push through, but I think we can
> do it.
>
>> Plan to periodically take a backup from your live database, and use that
>> re-init your slaves. MySQL replication is quite good, but I still have seen
>> some odd situations where replication can mess up. Since a full live diff
>> of the 2 servers (which would check all data and verify you're in sync)
>> isn't really feasible, I think it's safest to take backups from the prod
>> server on a somewhat-infrequent basis.
>
> I'll have to think about this one. I see your point and agree with it. I
> might instead tackle it via audits. I like audits for the potential of
> catching slow corruption. Might do both :).
>
> ciao,
>
> der.hans
> --
> # http://www.LuftHans.com/Classes http://www.TwoGeekTechs.com/
> # Director of Engineering, FonWallet Transaction Solutions, Inc.
> # Fairy Tale, n.: A horror story to prepare children for the newspapers.
> ---------------------------------------------------
> PLUG-discuss mailing list - PLUG-discuss@lists.plug.phoenix.az.us
> To subscribe, unsubscribe, or to change your mail settings:
> http://lists.PLUG.phoenix.az.us/mailman/listinfo/plug-discuss
>
---------------------------------------------------
PLUG-discuss mailing list -
PLUG-discuss@lists.plug.phoenix.az.us
To subscribe, unsubscribe, or to change your mail settings:
http://lists.PLUG.phoenix.az.us/mailman/listinfo/plug-discuss
(text/plain)