Thread: Backup of entire servers and roles/tablespaces
I'm just in the process of hacking pgAdmin to allow backup of entire servers or global info (roles and tablespaces). I was using the same mechanism as we use for pg_dump, however this doesn't work with pg_dumpall which doesn't support the -f option to redirect output to a file. Instead, I've been looking at separating the output and error streams from pg_dumpall, however this won't work either because pg_dumpall calls pg_dump to do much of it's work, and we cannot attach to those streams through pg_dumpall. I'm therefore toying with writing our own backup code to act as an equivalent to pg_dumpall, but would really prefer not to do this as the backup code is extremely important to get right of course. Thoughts or ideas? Regards, Dave.
Hi, On Fri, 2007-01-05 at 16:47 +0000, Dave Page wrote: > Thoughts or ideas? Isn't it a better solution to add -f to pg_dumpall? Regards, -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, ODBCng - http://www.commandprompt.com/
Attachment
Hi. > I'm just in the process of hacking pgAdmin to allow backup of entire > servers or global info (roles and tablespaces). I was using the same > mechanism as we use for pg_dump, however this doesn't work with > pg_dumpall which doesn't support the -f option to redirect output to a file. > > Instead, I've been looking at separating the output and error streams > from pg_dumpall, however this won't work either because pg_dumpall calls > pg_dump to do much of it's work, and we cannot attach to those streams > through pg_dumpall. > > I'm therefore toying with writing our own backup code to act as an > equivalent to pg_dumpall, but would really prefer not to do this as the > backup code is extremely important to get right of course. > > Thoughts or ideas? Yea, I think so. however, surely the importance exists for a user. I know that it has many problems with pgpass.conf.... Although only a specific user may be able to save, pgAdmin3 does not have preparations in it. probably, it is desirable. Regards, Hiroshi Saito
Devrim GUNDUZ a ecrit le 05/01/2007 17:55: > Hi, > On Fri, 2007-01-05 at 16:47 +0000, Dave Page wrote: > >> Thoughts or ideas? > > Isn't it a better solution to add -f to pg_dumpall? > Adding -f to pg_dumpall will only made it available for 8.3, isn't it ? or we will have to handle our own pg_dumpall source code with this new option. It seems a bit hard to me. -- Guillaume.
Guillaume Lelarge wrote: > Devrim GUNDUZ a ecrit le 05/01/2007 17:55: >> Hi, >> On Fri, 2007-01-05 at 16:47 +0000, Dave Page wrote: >> >>> Thoughts or ideas? >> >> Isn't it a better solution to add -f to pg_dumpall? >> > > Adding -f to pg_dumpall will only made it available for 8.3, isn't it ? > or we will have to handle our own pg_dumpall source code with this new > option. It seems a bit hard to me. Well, this code won't be released until 8.3 is anyway... I imagine it's not there already because when passed to pg_dump, the file will get overwritten every time. We'd need to add an append option to pg_dump as well. which seems a bit ugly. At first thought I figured there was more to it than that, but maybe not. I'll raise it on pgsql-hackers and see what objections are raised. Cheers, Dave.
Hi, On Fri, 2007-01-05 at 18:10 +0100, Guillaume Lelarge wrote: > > Isn't it a better solution to add -f to pg_dumpall? > > > > Adding -f to pg_dumpall will only made it available for 8.3, isn't > it ? Yes. But this will fix the source of the problem. Then pgadmin will say "I can dumpall to a file with PostgreSQL 8.3" That's all. Resolving this issue in pgadmin is the bad way, I think. Regards, -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, ODBCng - http://www.commandprompt.com/