From: | Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> |
---|---|
To: | Robert Haas <robertmhaas(at)gmail(dot)com> |
Cc: | Andres Freund <andres(at)anarazel(dot)de>, Noah Misch <noah(at)leadboat(dot)com>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org>, Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>, Ants Aasma <ants(dot)aasma(at)eesti(dot)ee>, Kevin Grittner <kgrittn(at)postgresql(dot)org>, Kevin Grittner <kgrittn(at)gmail(dot)com> |
Subject: | Re: old_snapshot_threshold allows heap:toast disagreement |
Date: | 2016-08-02 12:00:46 |
Message-ID: | CAA4eK1LMVg0K0vOaN3eK4KqQAU_7OLW=HFWGWUEYdzmJ8CJ6uQ@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Tue, Aug 2, 2016 at 9:10 AM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> On Sat, Jul 30, 2016 at 8:17 AM, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>> On Fri, Jul 29, 2016 at 1:10 AM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>>> On Wed, Jul 27, 2016 at 7:26 PM, Andres Freund <andres(at)anarazel(dot)de> wrote:
>>>
>>> New version attached.
>>
>> +static inline void
>> +InitToastSnapshot(Snapshot snapshot, XLogRecPtr lsn)
>> +{
>> + snapshot->satisfies = HeapTupleSatisfiesToast;
>> + snapshot->lsn = lsn;
>> +}
>>
>> Here, don't you need to initialize whenTaken as that is also used in
>> TestForOldSnapshot_impl() to report error "snapshot too old".
>
> Hmm, yeah. This is actually a bit confusing. We want the "oldest"
> snapshot, but there are three different notions of "oldest":
>
> 1. Smallest LSN.
> 2. Smallest whenTaken.
> 3. Smallest xmin.
>
> Which one do we use?
>
Which ever notion we choose, I think we should use the LSN and
whenTaken from the same snapshot. I think we can choose the snapshot
with smallest xmin and then use the LSN and whenTaken from it. I
think xmin comparison makes more sense because you are already
retrieving smallest xmin snapshot from the registered snapshots.
Whatever value we choose here, I think we can't guarantee that it will
be in sync with the value used for heap. So there is a chance that we
might spuriously raise "snapshot too old" error. I think we should
mentioned this as a caveat in docs [1] where we explain behaviour of
about old_snapshot_threshold.
--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com
From | Date | Subject | |
---|---|---|---|
Next Message | Etsuro Fujita | 2016-08-02 12:35:31 | Re: Oddity in EXPLAIN for foreign/custom join pushdown plans |
Previous Message | Robert Haas | 2016-08-02 11:45:21 | Re: [sqlsmith] Failed assertion in joinrels.c |