Thread: [HACKERS] pg_restore is broken on 9.2 version.
Commit c59a1a89035674c6efacc596d652528cebba37ec don't allow non-positive number
of jobs. Now on 9.2 number of jobs get assigned to opts->number_of_jobs. If
user don't specify any value -j then default value will be always 0. Which will
lead to the "invalid number of parallel jobs" error.
if (opts->number_of_jobs <= 0)
{
fprintf(stderr, _("%s: invalid number of parallel jobs\n"), progname);
exit(1);
}
Rushabh Lathia
Attachment
Greetings, * Rushabh Lathia (rushabh.lathia@gmail.com) wrote: > Commit c59a1a89035674c6efacc596d652528cebba37ec don't allow non-positive > number > of jobs. Now on 9.2 number of jobs get assigned to opts->number_of_jobs. If > user don't specify any value -j then default value will be always 0. Which > will > lead to the "invalid number of parallel jobs" error. > > if (opts->number_of_jobs <= 0) > { > fprintf(stderr, _("%s: invalid number of parallel jobs\n"), > progname); > exit(1); > } > > Please find attach patch to initialize default value for number of jobs to > 1. Ugh. This is what I get for thinking that our regression tests actually test even the basic things. That fix should go into NewRestoreOptions() where the other not-zero settings go. I'll fix it here in a few. Thanks! Stephen
* Stephen Frost (sfrost@snowman.net) wrote: > Greetings, > > * Rushabh Lathia (rushabh.lathia@gmail.com) wrote: > > Commit c59a1a89035674c6efacc596d652528cebba37ec don't allow non-positive > > number > > of jobs. Now on 9.2 number of jobs get assigned to opts->number_of_jobs. If > > user don't specify any value -j then default value will be always 0. Which > > will > > lead to the "invalid number of parallel jobs" error. > > > > if (opts->number_of_jobs <= 0) > > { > > fprintf(stderr, _("%s: invalid number of parallel jobs\n"), > > progname); > > exit(1); > > } > > > > Please find attach patch to initialize default value for number of jobs to > > 1. > > Ugh. This is what I get for thinking that our regression tests actually > test even the basic things. > > That fix should go into NewRestoreOptions() where the other not-zero > settings go. > > I'll fix it here in a few. Fix pushed. Thanks! Stephen