Please see the version 4.0 of HOT WIP patch posted on pgsql-patches.
I am having some trouble since afternoon posting the patch, tried
multiple times. So not sure if has made to -patches yet. If doesn't get
through on -patches this time also I would retry after few hours again.
Please read through the other important document for the time being.
ISTM that this version has one of the most radical changes included i.e.
the line pointer redirection.
The patch should apply on the current CVS HEAD and pass all regression
tests. I feel that the patch is now in a stable state and I would
really appreciate if community folks can take a look at it and
The VACUUM FULL is disabled right now. I am not aware of any design
issues for not supporting VACUUM FULL. I know its broken, but do
not want to hold back the patch while we fix it.
The most important change in this patch is the implementation of line
pointer redirection for handling dead root tuples. I have explained
the idea in the previous posts, but please see the attached write-up
for better illustration of the idea.
I would post some preliminary numbers in a separate email.
I would also post items where we need attention and discussion
and further work. I would welcome others to test the patch, for
correctness as well as performance.
pgsql-hackers by date
|Next:||From: Zeugswetter Andreas ADI SD||Date: 2007-03-01 14:03:39|
|Subject: Re: Revitalising VACUUM FULL for 8.3|
|Previous:||From: Jonathan Scher||Date: 2007-03-01 13:55:35|
|Subject: CLUSTER, using SHARE UPDATE EXCLUSIVE lock?|