Обсуждение: statistics not shown up
I start postgres using the -s parameter to track the number of hits to buffer. I also set both "stats_start_collector", "stats_block_level" to true in /usr/local/pgsql/data/postgresql.conf However, I get "(typeid = 700, len = 4, typmod = -1, byval = f)" after running each query. Apparently this is not manifesting buffer hit rate. Something I did wrong?
Seems that the guys @ pgfoundry are in problems but they do use postgresql ;)
An error occured in the logger. ERROR: could not extend relation "activity_log": No space left on device HINT: Check free disk space.
:(
On Sun, Sep 04, 2005 at 08:00:21PM -0400, huaxin zhang wrote: > I start postgres using the -s parameter to track the number of hits to buffer. I assume you're running postgres in single-user mode since that's where the -s option means "print statistics." Is that what you're doing? > I also set both "stats_start_collector", "stats_block_level" to true > in /usr/local/pgsql/data/postgresql.conf > > However, I get "(typeid = 700, len = 4, typmod = -1, byval = f)" after > running each query. Apparently this is not manifesting buffer hit rate. Try setting log_min_messages to debug1; that works for me in 8.0.3. -- Michael Fuhr
Nope, just moved the database server(s) to a faster server, and didn't realize that I hadn't cleared out enough disk space before doing so :( Already fixed, and 20+Gig of free space (and growing) to play with now) ... On Mon, 5 Sep 2005, Guido Barosio wrote: > Seems that the guys @ pgfoundry are in problems but they do use postgresql > ;) > An error occured in the logger. ERROR: could not extend relation > "activity_log": No space left on device HINT: Check free disk space. > :( > ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
I am using 8.0.3 One problem I found is that even if I specify to create a clustered index by "cluster idx_name on table_name", I still can not tell the optimizer that the index of "idx_name" is a clustered index with "index correlation" value as 1.0 (traced from the cost_index() in costsize.c) so where is the "index correlation" information stored and how can i update it? thanks Huaxin
huaxin zhang <uwcssa@gmail.com> writes: > One problem I found is that even if I specify to create a clustered index by > "cluster idx_name on table_name", I still can not tell the optimizer that the > index of "idx_name" is a clustered index with "index correlation" value as 1.0 > (traced from the cost_index() in costsize.c) The next ANALYZE should figure that out. regards, tom lane