[DRBD-user] oracle on drbd failed

Felix Frank ff at mpexnet.de
Wed Aug 29 15:00:32 CEST 2012

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

On 08/29/2012 02:51 PM, Mia Lueng wrote:
> I don't think so , Protocol A can still keep the data write order on
> secondary node just as on primary node.  It just link system crashed
> on primary node and resume when the node restart.

Well, a primary crash is handled rather gracefully even with protocol A.

The trouble is this: The Oracle process on your primary basically has no
idea what has and has not been written to the peer's disk, because the
kernel gives feedback only concerning writes to the primary's disk.

> 2012/8/29 Felix Frank <ff at mpexnet.de>:
>> > On 08/29/2012 03:16 AM, Mia Lueng wrote:
>>> >> Is it sure that  protocol A can not keep the oracle's data integration ?
>> >
>> > Protocol A cannot guarantee data integrity of *anything* in case of
>> > connectivity issues.
>> >
>> > Rule of thumb: When DRBD is backing your transactional database and you
>> > care not to lose transactions - use protocol C.
>> >
>> > HTH,
>> > Felix (a different one)

More information about the drbd-user mailing list