PITR- timeline query

Поиск
Список
Период
Сортировка
От Satya Prakash Tripathi
Тема PITR- timeline query
Дата
Msg-id 20060713151636.55608.qmail@web52012.mail.yahoo.com
обсуждение исходный текст
Ответы Re: PITR- timeline query  (Satya Prakash Tripathi <sptripathi78@yahoo.com>)
Список pgsql-admin
Hi ,

I am experimenting with PITR aspect of postgres8.0,
and I needed some answers on this subject.

I have following questions :

1)
At time t0, I created a base backup. (got a backup
marker file named as:
0000000800000002000000A8.005E61B4.backup)
(assume no previous base backups and recoveries...)

On time t2 I did a recovery till time t1 ( t0<t1<t2).
Now I see history file named 00000009.history. And
then onwards, the WAL filenames start with 00000009.
Is 00000009 here the new timeline ? If so is it safe
to deduce that from .history file that gets generated
after recovery ?

2)
Now say, I was happy with the above recovery and after
a few days, at time t3 I decided to revover my db back
in time t.

What timeline should I specify in recovery.conf if:
(or do I need to specify it in each case?)

  a)  t1 < t < t2
  b)  t2 < t < t3

IMO,  in case a) the timeline of base backup would be
used, so I wouldn't need to specify timeline. While in
case b) I need to specify this new
timeline(recovery_target_timeline=00000009).
Kindly confirm...

trying to put the scenario pictorially:

  t0                              t2
  |               t1              |
  |-------| ------ |---------- timeline: 00000008
                                   |
                                   |    timeline:
00000009

|----------------------|----

                              t3



TIA,
Satya



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

В списке pgsql-admin по дате отправления:

Предыдущее
От: "Hyatt, Gordon"
Дата:
Сообщение: Re: PostgreSQL 8.1.4 install failure on Win XP Home laptop
Следующее
От: "Rodrigo De Leon"
Дата:
Сообщение: Re: 8.1 Unique Index Issue/Bug???