On 09/15/2015 11:54 AM, Tom Lane wrote:
> Jan Wieck <jan@wi3ck.info> writes:
>> Would it be appropriate to use (Un)RegisterXactCallback() in case of
>> detecting excessive sequential scanning of that cache and remove all
>> invalid entries from it at that time?
> Another idea is that it's not the size of the cache that's the problem,
> it's the cost of finding the entries that need to be invalidated.
> So we could also consider adding list links that chain only the entries
> that are currently marked valid. If the list gets too long, we could mark
> them all invalid and thereby reset the list to nil. This doesn't do
> anything for the cache's space consumption, but that wasn't what you were
> worried about anyway was it?
That sounds like a workable solution to this edge case. I'll see how
that works.
Thanks, Jan
--
Jan Wieck
Senior Software Engineer
http://slony.info
