| From: | Marc Munro <marc(at)bloodnok(dot)com> |
|---|---|
| To: | pgsql-general(at)postgresql(dot)org |
| Subject: | Transaction id wraparound questions |
| Date: | 2005-07-06 18:45:04 |
| Message-ID: | 1120675504.24809.16.camel@bloodnok.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
It seems that we have not been vacuuming our production database
properly. We have been explicitly vacuuming all tables individually but
have not vacuumed the entire database.
A recent vacuum of the entire database gave us the dreaded
"You may have already suffered transaction-wraparound data loss."
warning.
We have so far encountered no problems but I am wondering about the
safest course of action right now. We cannot easily take an outage to
perform a full dump and restore.
Questions:
1) What is likely to happen if we encounter transaction id wraparound?
2) Will a full database vacuum fix the problem?
3) Can it make things worse?
4) Other than dump and restore, what options do we have?
Information:
This query:
select datname, datvacuumxid, datfrozenxid from pg_database;
returns this:
"Production",1173213507,2246955329
"template1", 938856359, 938856359
"template0", 427, 427
All responses welcomed.
__
Marc Munro
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tom Lane | 2005-07-06 19:04:44 | Re: Transaction id wraparound questions |
| Previous Message | Mike Fahrenkrog | 2005-07-06 18:35:50 | Scripting Functions to Separate Files |