On Tue, Feb 24, 2009 at 11:42 AM, Greg Smith
<gsmith@gregsmith.com> wrote:
Right, the useful thing to do in this case is to take a look at how big all the relations (tables, indexes) involved are at each of the steps in the process. The script at http://wiki.postgresql.org/wiki/Disk_Usage will show you that. That will give some feedback on whether the vacuum/reindex methodology is really doing what you expect, and it will also let you compare the size of the table/index with how much RAM is in the system.
taken before the ~7000 rows were entered.
relation | size
-------------------------------+--------
public.mdl_log | 595 MB
public.mdl_forum_posts | 375 MB
public.mdl_log_coumodact_ix | 197 MB
public.mdl_user | 191 MB
public.mdl_cache_text | 162 MB
public.mdl_log_usecou_ix | 137 MB
public.mdl_log_act_ix | 119 MB
public.mdl_log_cmi_ix | 97 MB
public.mdl_log_tim_ix | 97 MB
public.mdl_log_id_pk | 97 MB
public.mdl_question_states | 48 MB
public.mdl_stats_user_daily | 48 MB
public.mdl_hotpot_responses | 47 MB
public.mdl_register_downloads | 45 MB
public.mdl_message_read | 37 MB
public.mdl_course_display | 37 MB
public.mdl_stats_user_weekly | 31 MB
public.mdl_mnet_log | 27 MB
public.mdl_user_ema_ix | 26 MB
public.mdl_regidown_url_ix | 23 MB
(20 rows)