[DRBD-user] Database integrity

Corey Edwards tensai at zmonkey.org
Wed May 10 21:36:46 CEST 2006

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


On Wed, 2006-05-10 at 21:49 +0300, Alex wrote:
>    Hello gentlemen,
>    We solved try to use DRBD with Heartbeat for make more reliable
> work of PostgreSQL database. But after introduction with most part of
> documentation i hasn't found straight answer on one question.
>    Can DRBD guarantee integrity of database (all tables, indexes, cache
> and other parts) after fail of active node and switching in active mode
> passive node with start of database?

Using protocol C, DRBD will be no more risky than using a filesystem on
local disk. With protocol B you add a slight risk that in the event of
an outage on the master, a write which was occurring on the slave could
fail but still be recorded on the master. With protocol A that window of
opportunity is greater and generally shouldn't be used unless you have a
high latency network and know what you're doing.

In all cases, the master will only record a filesystem transaction after
it reaches local disk, so you will always be better off with DRBD. I
have had disks crash w/ DRBD and suffered much less than those without.
We even had both disks in a RAID 1 die, but DRBD saved our butt by
feeding the master with data from the slave. What an ingenious idea.

Bottom line, use DRBD. Your database will thank you for it.

Corey

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: This is a digitally signed message part
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20060510/566c462d/attachment.pgp>


More information about the drbd-user mailing list