| From: | Michael Paquier <michael(dot)paquier(at)gmail(dot)com> |
|---|---|
| To: | Andreas Seltenreich <seltenreich(at)gmx(dot)de> |
| Cc: | PostgreSQL mailing lists <pgsql-hackers(at)postgresql(dot)org>, Magnus Hagander <magnus(at)hagander(dot)net> |
| Subject: | Re: [sqlsmith] FailedAssertion("!(XLogCtl->Insert.exclusiveBackup)", File: "xlog.c", Line: 10200) |
| Date: | 2016-08-04 07:38:40 |
| Message-ID: | CAB7nPqRf7=dx8Ufk1svdGHzz1LO8XN=sZj_X7c08eLc9h0h91Q@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Thu, Aug 4, 2016 at 4:03 PM, Michael Paquier
<michael(dot)paquier(at)gmail(dot)com> wrote:
> This window exists as well on back-branches btw, this is not new to
> 9.6. Magnus, what do you think?
Actually, a completely water-proof solution is to use an in-memory
flag proper to exclusive backups during the time pg_start_backup() is
called, at the cost of taking WALInsertLockAcquireExclusive once again
at the end of do_pg_start_backup(), but it seems to me that it is an
acceptable cost because pg_start_backup is not a hot code path. Using
a separate flag has the advantage to provide to the user a better
error message when pg_stop_backup is used while pg_start_backup is
running as well.
Andreas, with the patch attached is the assertion still triggered?
Thoughts from others are welcome as well.
--
Michael
| Attachment | Content-Type | Size |
|---|---|---|
| base-backup-crash-v2.patch | text/x-diff | 2.7 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tsunakawa, Takayuki | 2016-08-04 07:41:48 | [RFC] Change the default of update_process_title to off |
| Previous Message | Andres Freund | 2016-08-04 07:24:47 | Re: Reviewing freeze map code |