Good afternoon folks.
I've been playing around with some vacuumdb options this week as part of post-upgrade testing, in particular with parallel (--jobs=N). I noticed that vacuumdb --all will only work on one database at a time. This means that as it winds down to the last few tables in a particular database, workers will be idle. For example, I ran with 8 jobs, but while 1 job runs on the last big table of a particular database, the other 7 are doing nothing.
It would be great if those idle jobs would be able to move on to the next database to get a head start on the tables there. When you have big tables in multiple databases in the cluster, it can make a big difference.
Also, running analyze_new_cluster.sh mentions this:
If you would like default statistics as quickly as possible, cancel
this script and run:
"/usr/pgsql-9.6/bin/vacuumdb" --all --analyze-only
I wonder if it wouldn't be a bad idea to also mention the --jobs=N parameter option in that blurb. Yes it's in the --help text but it wouldn't be bad to highlight its availability.
