| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Michael Paquier <michael(at)paquier(dot)xyz> |
| Cc: | Euler Taveira <euler(at)timbira(dot)com(dot)br>, Tatsuo Ishii <ishii(at)sraoss(dot)co(dot)jp>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: Adding Unix domain socket path and port to pg_stat_get_wal_senders() |
| Date: | 2019-04-12 13:43:34 |
| Message-ID: | 15919.1555076614@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Michael Paquier <michael(at)paquier(dot)xyz> writes:
> On Thu, Apr 11, 2019 at 10:19:01PM -0300, Euler Taveira wrote:
>> Socket has different semantic from TCP/UDP. We can't add socket
>> information into client_addr unless we are prepared to break this view
>> (client_addr has type inet and it would be necessary to change it to
>> text). It could break a lot of applications.
Agreed.
> client_addr does not seem the right place to store this information,
> and it is already documented for years that NULL is used when using a
> Unix socket. But I think that we could change *client_hostname* so as
> the path name is reported instead of NULL when connecting through a
> Unix domain socket, and there is no need to switch the field type for
> that.
That seems like a hack, and I think it could still break apps that
are expecting particular semantics for that field. Why not add a
new column instead?
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Julien Rouhaud | 2019-04-12 13:54:23 | Re: Checksum errors in pg_stat_database |
| Previous Message | Daniel Gustafsson | 2019-04-12 13:36:59 | Re: PostgreSQL pollutes the file system |