Re: Proposal: Log inability to lock pages during vacuum

On 11/10/14, 11:28 AM, Alvaro Herrera wrote:
> Jim Nasby wrote:
>> On 11/7/14, 8:21 PM, Robert Haas wrote:
>>> On Thu, Nov 6, 2014 at 8:03 PM, Jim Nasby <Jim(dot)Nasby(at)bluetreble(dot)com> wrote:
>>>> The problem right now is there's no way to actually obtain evidence that
>>>> this is (or isn't) something to worry about, because we just silently skip
>>>> pages. If we had any kind of tracking on this we could stop guessing. :(
>>>
>>> I could see logging it, but I agree with Andres and Alvaro that the
>>> odds are strongly against there being any actual problem here.
>>
>> I'm fine with that. Any other objections? Andres?
>
> If what we want is to quantify the extent of the issue, would it be more
> convenient to save counters to pgstat? Vacuum already sends pgstat
> messages, so there's no additional traffic there.

IMHO that would be ideal, but I think Tom was leery of using more space for every table. If we go this route, I'm guessing we should only log pages we skip, and not log pages we had to wait for the lock on (in the case of a freeze). Also, should we still eroprt this even if we are putting it in stats?

Is there a way to avoid duplicating the entire eroprt call? I see I could call errstart & friends manually, but currently that's only done in elog.c.
--
Jim Nasby, Data Architect, Blue Treble Consulting
Data in Trouble? Get it in Treble! http://BlueTreble.com