From: Igor Neyman <ineyman@xxxxxxxxxxxxxx>
To: Stephen Frost <sfrost@xxxxxxxxxxx>; Patrick Dung <patrick_dkt@xxxxxxxxxxxx>
Cc: "pgsql-general@xxxxxxxxxxxxxx" <pgsql-general@xxxxxxxxxxxxxx>; Ivan Voras <ivoras@xxxxxxxxxxx>; Tom Lane <tgl@xxxxxxxxxxxxx>
Sent: Saturday, September 14, 2013 2:14 AM
Subject: RE: Major upgrade of PostgreSQL and MySQL
Hi Igor,
>> -----Original Message-----
>> From: pgsql-general-owner@xxxxxxxxxxxxxx [mailto:pgsql-general-
>> owner@xxxxxxxxxxxxxx] On Behalf Of Stephen Frost
>> Sent: Friday, September 13, 2013 2:06 PM
>> To: Patrick Dung
>> Cc: pgsql-general@xxxxxxxxxxxxxx; Ivan Voras; Tom Lane
>> Subject: Re: Major upgrade of PostgreSQL and MySQL
>>
>>
>> > 3. But the way, if users is using Windows, is the link option still works?
>>
>> Don't think so, but not sure. pg_upgrade could be made to work in a truely
>> "in-place" method if there's demand for it and someone wants to work on it.
>> It'd clearly be a bit more *dangerous*, of course..
>>
>> Thanks,
>>
>> Stephen
>
>Like I said in the other message, actually in-place upgrade using symbolic links work quite fine under Windows.
>I tested it carefully many times, and used it even more upgrading production systems.
>I don't feel it's *dangerous*, especially considering that my whole upgrade process always starts with backing up existing cluster.
>
For Windows, is it using symbolic links or hard links for the upgrade?
If symbolic links is used, would users have difficultly when deleting the old cluster?
Thanks,
Patrick
