| From: | Tomas Vondra <tomas(dot)vondra(at)enterprisedb(dot)com> |
|---|---|
| To: | Heikki Linnakangas <hlinnaka(at)iki(dot)fi>, Jelte Fennema-Nio <postgres(at)jeltef(dot)nl> |
| Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>, Thomas Munro <thomas(dot)munro(at)gmail(dot)com>, Robert Haas <robertmhaas(at)gmail(dot)com> |
| Subject: | Re: Make query cancellation keys longer |
| Date: | 2024-07-04 12:43:20 |
| Message-ID: | 328b01a4-b2c0-40da-ab14-f7f005d4e0e3@enterprisedb.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
I don't have any immediate feedback regarding this patch, but I'm
wondering about one thing related to cancellations - we talk cancelling
a query, but we really target a PID (or a particular backend, no matter
how we identify it).
I occasionally want to only cancel a particular query, but I don't think
that's really possible - the query may complete meanwhile, and the
backend may even get used for a different user connection (e.g. with a
connection pool)? Or am I missing something important?
Anyway, I wonder if making the cancellation key longer (or variable
length) might be useful for this too - it would allow including some
sort of optional "query ID" in the request, not just the PID. (Maybe
st_xact_start_timestamp would work?)
Obviously, that's not up to this patch, but it's somewhat related.
regards
--
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Yao Wang | 2024-07-04 12:45:33 | Re: 回复: An implementation of multi-key sort |
| Previous Message | Erik Wienhold | 2024-07-04 12:34:35 | Re: Underscore in positional parameters? |