| From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
|---|---|
| To: | Andres Freund <andres(at)anarazel(dot)de> |
| Cc: | Joshua Chamberlain <josh(at)zephyri(dot)co>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: CREATE TABLE with parallel workers, 10.0? |
| Date: | 2017-02-16 14:16:00 |
| Message-ID: | CA+TgmobRbV=Q5jY2M9UzXh2o2qR=935fPeOJ-Y3Y5Uxvr-Q3oA@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Wed, Feb 15, 2017 at 8:13 PM, Andres Freund <andres(at)anarazel(dot)de> wrote:
> I don't think general INSERTs are safe, if you consider unique indexes
> and foreign keys (both setting xmax in the simple cases and multixacts
> are likely to be problematic).
There's no real problem with setting xmax or creating multixacts - I
think - but there's definitely a problem if an INSERT can lead to the
creation of a new combo CID, because we have no way at present of
transmitting the new combo CID mapping to the workers, and if a worker
sees a combo CID for which it doesn't have a mapping, the world blows
up. Do you think an insert can trigger combo CID creation?
(Of course, now that we've got DSA, it wouldn't be nearly as hard to
fix the combo CID synchronization problem.)
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Amit Kapila | 2017-02-16 14:19:39 | Re: Possible issue with expanded object infrastructure on Postgres 9.6.1 |
| Previous Message | Robert Haas | 2017-02-16 13:58:59 | Re: Should we cacheline align PGXACT? |