-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

I was just debugging the database of some old mmbase installations we 
have running... and then I noticed something very interesting: the 
install_mmservers table in the postgresql database turns out to have a 
size of 50Mb on disk! even though it should only contain 1 row (well, 
in some occasions a few more, which is a known bug in older mmbase I 
think) and we run a FULL VACUUM on the database every night...
If I do a 'select count(*)' on mmservers it takes several seconds 
before it finally tells me there are only 8 rows in it.

My theory is the following: earlier releases of mmbase did an update on 
mmservers every *other second*, which, considering the way postgresql 
handles data, builds up a lot of extra data in the structure... since 
mmbase is 'hammering' mmservers 24/7, VACUUM commands will probably 
never succeed on that table... so the longer you are using mmbase in 
postgresql, the more data it is going to waist... and will probably 
waist a lot of performance over time.

my question is: is there an easy way I can stop older mmbase 
installations to update mmservers 24/7 without having direct access to 
the sourcecode of that release? Maybe I could just drop & recreate the 
table once a day :)

and I'm wondering how this affects the current versions of mmbase (I 
know mmservers is updated less frequently but will this still affect 
the postgresql VACUUM commands?)

Regards,

Ricardo.

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.0 (Build 288) Beta

iQA/AwUBPbGBS15Af8OSarqUEQLTVACePH3GuapiKpz57lqrXOFuQggPl0IAn2V2
W/6+b0FaqlBGCgrNfrkSzhA+
=qAMJ
-----END PGP SIGNATURE-----


Reply via email to