On 6/10/2010 4:28 PM, Scott Marlowe wrote:
> On Thu, Jun 10, 2010 at 3:22 PM, Andy Colson<andy@squeakycode.net> wrote:
>> On 6/10/2010 3:25 PM, Aleksey Tsalolikhin wrote:
>>>
>>> PostgreSQL 8.4.2 response time slows during pg_dump of a 14 GB database.
>>>
>>> Specifically, I am monitoring how long it takes to login to our Web
>>> app - usually under 2 seconds, but when pg_dump is running, that
>>> changes to 2 - 18 seconds.
>>>
>>> Is there any way I can keep my response fast but still get my pg_dump?
>>> Or should I move my pg_dump to a Slony-I slave?
>>>
>>> Best,
>>> -at
>>>
>>
>> It depends:
>> if you are cpu bound, you could try nice.
>> if you are network bound then dont pg_dump across the network
>> if you are io bound, pg_dump to a different drive, or pg_dump across
>> the network.
>>
>>
>> Isnt the slave a backup anyway? Do you need another dump?
>
> Replication does not equal backup. For instance how would you recover
> from "delete from table" with no where clause with replication?
Ah, thats easy.. you just have to run faster than the packet, back to
the room with the slave it in, and unplug the network... :-)
but yeah, good point.
-Andy