Strange VACUUM behaviour

Hi

We started a VACUUM (not a VACUUM FULL) on one of your postgres 7.4.9
databases a few days ago. It's still running yet, and says the
folloing about once per second:

INFO: index "pg_toast_2144146_index" now contains 1971674 row versions
in 10018 pages
DETAIL: 4 index row versions were removed.
2489 index pages have been deleted, 0 are currently reusable.

The number of row versions decreases by 4 each time the message is logged.

The file belonging to pg_toast_2144146_index has about 80MB,
for pg_toast_2144146 there are 6 files, five of them are
1GB, the last one is about 5MB in size. The "original" relation
(the one that references pg_toast_2144146 in it's reltoastrelid field)
has one datafile of 11MB.

The "original" relation is called image, and is defined the following:
Table "public.image"
Column | Type | Modifiers
---------------+------------------------+-----------
id | bigint | not null
image_code_id | bigint |
mandant_id | bigint |
name | text |
dat | text |
mime | text |
size | bigint |
md5 | bytea |
path | text |
copyright | character varying(255) |
Indexes:
"image_pkey" primary key, btree (id)
"i_image_id" btree (id)
Triggers:
_gti_denyaccess_17 BEFORE INSERT OR DELETE OR UPDATE ON image FOR
EACH ROW EXECUTE PROCEDURE _gti.denyaccess('_gti')

The table is part of a slony tableset, which is subscribed on this database.

Is there a reason that this vacuum takes so long? Maybe some lock
contention because slony replicates into this table?

greetings, Florian Pflug