Bruce Momjian <bruce@momjian.us> wrote:
> On Wed, Sep 10, 2014 at 02:07:36PM -0700, Kevin Grittner wrote:
>> Bruce Momjian <bruce@momjian.us> wrote:
>>> On Sat, Sep 6, 2014 at 11:07:43AM -0400, Bruce Momjian wrote:
>>>> Here is a patch which implements the warning during CREATE INDEX
> ...
>>>> HASH. If WAL-logging of hash indexes is ever implemented, we can remove
>>>> this warning.
>>>
>>> Applied, though I used the term "streaming standbys" to match our docs.
>>
>> Hmm. The wording of the warning doesn't seem to really indicate
>> the full scope of the limitation. Any a standby (warm or hot)
>> maintained by WAL file copying would also be affected (i.e.,
>> streaming replication as the WAL delivery mechanism is irrelevant),
>> and you also have problems after a database crash or PANIC. I'm
>> not sure how to state that concisely, though.
>
> I am open to improved wording. :-)
Instead of:
hash indexes are not WAL-logged so they are not crash-safe and cannot be used on streaming standbys
how about?:
hash indexes are not WAL-logged and thus are not crash-safe and cannot be used on standby servers
That seems consistent with the terminology used here:
http://www.postgresql.org/docs/current/interactive/warm-standby.html
--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company