Thread: problem in restoring database when connected via named socket
Hi all, I am connected to a local postgresql server via named pipe instead of tcp/ip socket, i.e., I did not specify the hostname while creating the connection. No, when I try to restore a database, the command fail since the "hostname" argument is missing. This is the error message I get: /usr/bin/pg_restore -i -h -p 5439 -U giuseppe -d proteus -v "/home/giuseppe/Desktop/migrazione saronno/saronno-proteus.pgdump" pg_restore: [archiviatore] could not open input file "5439": No such file or directory the problem being that pg_restore use "-p" as hostname since it is just after "-h" option, then "4539" is used as file name since it is the first non option argument. Of course, I imported my dump using command line pg_restore. Using the very same line, without "-h". Is this a known problem of pgadmin? Should I file it somewhere? Bye, Giuseppe
Hi Giuseppe, Le 12/01/2010 10:34, Giuseppe Sacco a écrit : > [...] > I am connected to a local postgresql server via named pipe instead of > tcp/ip socket, i.e., I did not specify the hostname while creating the > connection. > > No, when I try to restore a database, the command fail since the > "hostname" argument is missing. This is the error message I get: > > /usr/bin/pg_restore -i -h -p 5439 -U giuseppe -d proteus -v > "/home/giuseppe/Desktop/migrazione saronno/saronno-proteus.pgdump" > pg_restore: [archiviatore] could not open input file "5439": No such > file or directory > > the problem being that pg_restore use "-p" as hostname since it is just > after "-h" option, then "4539" is used as file name since it is the > first non option argument. > > Of course, I imported my dump using command line pg_restore. Using the > very same line, without "-h". > > Is this a known problem of pgadmin? Should I file it somewhere? > You shouldn't have this issue because we check if the server name is empty or not before adding the -h command line option. AFAICT, this is fixed since "Wed Mar 19 11:28:37 2008". So I assume you have a rather old release. Try on a current one (current being 1.10.1 right now). -- Guillaume.http://www.postgresqlfr.orghttp://dalibo.com
Hi Guillaume, Il giorno mar, 12/01/2010 alle 10.47 +0100, Guillaume Lelarge ha scritto: > Hi Giuseppe, [...] > You shouldn't have this issue because we check if the server name is > empty or not before adding the -h command line option. AFAICT, this is > fixed since "Wed Mar 19 11:28:37 2008". So I assume you have a rather > old release. Try on a current one (current being 1.10.1 right now). Right, I am using 1.9. Yesterday I tried to update source from repository, but I did not manage to compile it because of a link error. I'll try again later on. Thanks, Giuseppe
Le 12/01/2010 11:08, Giuseppe Sacco a écrit : > Hi Guillaume, > > Il giorno mar, 12/01/2010 alle 10.47 +0100, Guillaume Lelarge ha > scritto: >> Hi Giuseppe, > [...] >> You shouldn't have this issue because we check if the server name is >> empty or not before adding the -h command line option. AFAICT, this is >> fixed since "Wed Mar 19 11:28:37 2008". So I assume you have a rather >> old release. Try on a current one (current being 1.10.1 right now). > > Right, I am using 1.9. Yesterday I tried to update source from > repository, but I did not manage to compile it because of a link error. > I'll try again later on. > Did you re-run ./bootstrap? if not, try to do so. -- Guillaume.http://www.postgresqlfr.orghttp://dalibo.com
Il giorno mar, 12/01/2010 alle 11.21 +0100, Guillaume Lelarge ha scritto: > Did you re-run ./bootstrap? if not, try to do so. I didn't. That fixed the problem. Thanks again, Giuseppe