| From: | Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> |
|---|---|
| To: | Petr Jelinek <petr(dot)jelinek(at)2ndquadrant(dot)com> |
| Cc: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, pgsql-hackers(at)postgreSQL(dot)org, Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com> |
| Subject: | Re: More race conditions in logical replication |
| Date: | 2017-07-12 22:48:28 |
| Message-ID: | 20170712224828.iphg5ttm2h6egm3d@alvherre.pgsql |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Petr Jelinek wrote:
> So best idea I could come up with is to make use of the new condition
> variable API. That lets us wait for variable which can be set per slot.
Here's a cleaned up version of that patch, which I intend to get in the
tree tomorrow. I verified that this allows the subscription tests to
pass with Tom's sleep additions.
I noticed one point where we're reading the active_pid without locking;
marked it with an XXX comment. Not yet sure whether this is a bug or
not.
I noticed something funny in CREATE_REPLICATION; apparently we first
create a slot and set it active, then we activate it by name. With the
current coding it seems to work fine, because we're careful to make
activation idempotent, but it looks convoluted. I don't think this is
new, though.
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
| Attachment | Content-Type | Size |
|---|---|---|
| v3-0001-Wait-for-slot-to-become-free-in-when-dropping-it.patch | text/plain | 9.4 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Greg Stark | 2017-07-12 22:55:56 | Re: building libpq.a static library |
| Previous Message | Dean Rasheed | 2017-07-12 22:23:58 | Re: New partitioning - some feedback |