| From: | Fujii Masao <masao(dot)fujii(at)gmail(dot)com> |
|---|---|
| To: | PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Streaming Replication and CopyOutResponse message |
| Date: | 2010-03-26 09:30:32 |
| Message-ID: | 3f0b79eb1003260230q38be9b84m17aa79729942296b@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
At the end of the handshake for replication, a CopyOutResponse
message might not arrive at the standby for a while if there is
no WAL record to send. Also walreceiver would get stuck until
that message has arrived. This is not a big problem, but should
be fixed.
The cause is that walsender hasn't called pq_flush() there.
The attached patch changes walsender so as to call pg_flush()
immediately when sending that message. Is it worth applying
the patch?
Regards,
--
Fujii Masao
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center
| Attachment | Content-Type | Size |
|---|---|---|
| walsender_pg_flush_v1.patch | application/octet-stream | 341 bytes |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Marko Tiikkaja | 2010-03-26 09:44:05 | Re: Proposal: access control jails (and introduction as aspiring GSoC student) |
| Previous Message | Albe Laurenz | 2010-03-26 08:03:57 | Re: Ask help for putting SP-Gist into postgresql |