| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> | 
|---|---|
| To: | Victor Yegorov <viy(at)pirmabanka(dot)lv> | 
| Cc: | pgsql-admin(at)postgresql(dot)org | 
| Subject: | Re: DB archiving | 
| Date: | 2003-03-17 15:53:23 | 
| Message-ID: | 6522.1047916403@sss.pgh.pa.us | 
| Views: | Whole Thread | Raw Message | Download mbox | Resend email | 
| Thread: | |
| Lists: | pgsql-admin | 
Victor Yegorov <viy(at)pirmabanka(dot)lv> writes:
> One more thing. I still cannot find any info on HOW pg_dump locks data it
> is going to backup.
It doesn't.  Well, it does take a reader's lock (AccessShareLock) on
each table it intends to back up, but that's essentially just preventing
the table from being dropped completely.  The individual data rows need
not be locked.  pg_dump does the whole dump as a single transaction, so
it can rely on MVCC semantics to give it a consistent picture of the
data in every table.  Basically, it doesn't "see" changes that commit
after its transaction starts.  See the discussion of MVCC.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Matt Clark | 2003-03-17 15:56:13 | Re: recomended plataform | 
| Previous Message | Sidar Lopez Cruz | 2003-03-17 15:50:54 | recomended plataform |