Re: [HACKERS] 6.1 pg_dump core dump
От | bower@image.Kodak.COM (J Christopher Bower) |
---|---|
Тема | Re: [HACKERS] 6.1 pg_dump core dump |
Дата | |
Msg-id | 199808031304.JAA17648@coyote.image.Kodak.COM обсуждение исходный текст |
Список | pgsql-hackers |
>> bower@image.Kodak.COM (J Christopher Bower) writes: > I decided to debug the 6.1 pg_dump source and determined that the core dump > occured dumping tblinfo cleanup. Specifically when structures allocated for > sequence tables are being cleaned up. >> It looks like this bug has been fixed in a different way in the current >> pg_dump sources (clearTableInfo now knows that sequences don't have the >> standard attributes). It also looks like pg_dump has changed enough >> since 6.1 that any patches wouldn't be easily transferred back and >> forth anyway. >> You might be able to run the current pg_dump against your 6.1 database >> by recompiling the current pg_dump.c/.h/common.c atop the 6.1 libpq. >> That should cure the protocol incompatibility. However, pg_dump is >> friendly enough with the system table layouts that I fear it might not >> work with an old database anyway. >> Probably your best bet is just to go ahead and use your patched pg_dump >> to extract data from your old database. >> regards, tom lane Tom, Thanks for your help it looks like I'll proceed with my modified version of pg_dump. Compiling pg_dump 6.3.2 with the 6.1 libpq was successfull. This pg_dump could connect to the database but failed when it attempted to perform a select from pg_class in the 6.1 database and expected the attribute relcheck. Apparently pg_class has been restructure somewhere between 6.1 and 6.3.2. I'm confused because the documentation that I read indicates that the 6.3.2 pg_dump should be able to dump the 6.1 database. But appaently this cannot be done, atleast with my 6.1 database. Again thanks for your help. Chris Bower
В списке pgsql-hackers по дате отправления: