Yes it dramatically decreases the transaction flow. This function has not been implemented at all for performance but for analysis purposes. I used it mainly to have a look at state files size in pg_twophase for transactions that are prepared but not committed.
Regards On Fri, Aug 7, 2009 at 12:55 PM, Itagaki Takahiro < itagaki.takah...@oss.ntt.co.jp> wrote: > > Michael Paquier <michael.paqu...@gmail.com> wrote: > > > > Here is a short patch implementing a new feature in pgbench so as to > allow > > > shell commands to be launched in a transaction file of pgbench. > > > \shell ls ~/pg_twophase; > > +1 for \shell command itself, but does the performance fit for your > purpose? > Spawning a new process is not so cheap, no? > > -1 for -P option because it is too narrow purpose and 'ls' and '/tmp/' > is not portable. We don't need to include your workload because you can > use -f FILENAME to run your benchmark script. > > Regards, > --- > ITAGAKI Takahiro > NTT Open Source Software Center > > > -- Michael Paquier NTT OSSC
