Обсуждение: vacuum and postgresql.conf settings
I am not even sure this is a problem. INFO: "data_import_customer2": scanned 3000 of 5033 pages, containing 84905 live rows and 0 dead rows; 3000 rows in sample, 142442 estimated total rows Do I have some setting wrong?
Walter Vaughan writes: > I am not even sure this is a problem. > INFO: "data_import_customer2": scanned 3000 of 5033 pages, containing 84905 > live rows and 0 dead rows; 3000 rows in sample, 142442 estimated total rows I don't believe it is a problem. I think it is just normal output of the vacuum. What made you think there could be a problem?
Francisco Reyes wrote: > Walter Vaughan writes: >> I am not even sure this is a problem. >> INFO: "data_import_customer2": scanned 3000 of 5033 pages, containing >> 84905 live rows and 0 dead rows; 3000 rows in sample, 142442 estimated >> total rows > > I don't believe it is a problem. > I think it is just normal output of the vacuum. > What made you think there could be a problem? Well, why did it only scan 3000 pages out of 5033? It decided it didn't need to? It took a sample and decided it was gonna be okay?
Walter Vaughan writes: > Well, why did it only scan 3000 pages out of 5033? It decided it didn't need to? > It took a sample and decided it was gonna be okay? I don't know. Hopefully others with more knowledge can answer. I think it could be because the pages did not need to be looked at. Perhaps the database tracks whether a page has deleted information and those are the ones that vacuum looks into.
Walter Vaughan wrote: > Francisco Reyes wrote: > > >Walter Vaughan writes: > >>I am not even sure this is a problem. > >>INFO: "data_import_customer2": scanned 3000 of 5033 pages, containing > >>84905 live rows and 0 dead rows; 3000 rows in sample, 142442 estimated > >>total rows > > > >I don't believe it is a problem. > >I think it is just normal output of the vacuum. > >What made you think there could be a problem? > > Well, why did it only scan 3000 pages out of 5033? It decided it didn't > need to? It took a sample and decided it was gonna be okay? These lines are for the ANALYZE step. It doesn't need to scan all the pages. It picks 3000 at random for computing stats. The VACUUM step (which happens before ANALYZE) does walk all pages. -- Alvaro Herrera Valdivia, Chile ICBM: S 39º 49' 18.1", W 73º 13' 56.4" "Ni aun el genio muy grande llegaría muy lejos si tuviera que sacarlo todo de su propio interior" (Goethe)