Feel free to kill pg_autovacuum and try different parameters as much as you like, killing pg_autovacuum won't hurt anything. You should note that the current implementation doesn't keep any history of information, so when you restart pg_autovacuum, it sets it's update counts to 0. Matthew On Thu, 2004-05-20 at 10:12, Carlos Oliva wrote: > Thank you for your reply Matthew. Do you know if I can stop auto-vacuum, > and re-start it with different parameters as the uploading continues? I do > not know if my killing the auto-vacuum process would terminate my > application also. > > I want to start auto-vacuum with different parameters to tune the vacuuming > to my application. I would like to vacuum often as the data is being > uploaded (inserts mainly) because I want to maintain a high speed of upload > > Thanks again. > > -----Original Message----- > From: pgsql-general-owner@postgresql.org > [mailto:pgsql-general-owner@postgresql.org] On Behalf Of Matthew T. O'Connor > Sent: Thursday, May 20, 2004 12:54 AM > To: Carlos > Cc: 'pgsql-general@postgresql.org' > Subject: Re: Settings for autovacuum for batch uploading of data? > > More important than sleep value is the vacuum threshold. If for example > you are expecting to import 5000 rows of data, you can tell > pg_autovacuum to vacuum after every 5000 changes with like > > pg_autovacuum -v 5000 -V 0 > > Matthew > > > On Wed, 2004-05-19 at 17:09, Carlos wrote: > > Hello Forum, > > > > I would appreciate it if you could recommend settings to use > > auto-vacuum in my version 7.4 database. I am uploading several > > thousands records in the database at a rate of ~1 second per record > > (the data is uploading from a different server). I have noticed that > > the performance of the uploading is improved if I vacuum the database > > every so often so I decided to take advantage of auto-vacuuming to > > maintain the best performance. Since I am uploading several thousands > > records, I set autovacuum to 50 minutes with -s 3000 in the hope that > > a great deal of data would be committed before a regular vacuum would > > take place and that vacuuming would interfere the least possible with > > the uploading. > > > > > > > > Thanks in advance for your recommendations. > > > > > > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org > ---------------------------(end of broadcast)--------------------------- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
