![]() | |
![]() |
| | Thread Tools | Display Modes |
#1
| |||
| |||
|
#2
| |||
| |||
|
|
creating template1 database in /pg_database/database/base/1... ok initializing pg_shadow... ok enabling unlimited row size for system tables... ok initializing pg_depend... ok creating system views... ERROR: did not find '}' at end of input node |
|
I think this problem is related to HP-UX 11.00 because I test my upgrade procedure on an HP-UX 10.20 box, same database, same source and same destination version w/o any trouble. |
#3
| |||
| |||
|
|
creating system views... ERROR: did not find '}' at end of input node Hmm. This would seem to indicate a mismatch between the nodeToString and stringToNode routines (outfuncs.c and readfuncs.c), which is something we don't normally see unless someone's messed up their build. Try a full recompile. |
|
You're not using that infamous CONNECT BY patch, are you? |
#4
| |||
| |||
|
|
--- Tom Lane <tgl (AT) sss (DOT) pgh.pa.us> a écrit*: creating system views... ERROR: did not find '}' at end of input node Hmm. This would seem to indicate a mismatch between the nodeToString and stringToNode routines (outfuncs.c and readfuncs.c), which is something we don't normally see unless someone's messed up their build. Try a full recompile. |
#5
| |||
| |||
|
|
I do a full recompile (I have removed the directory and restart from the tarball) => Same result :-( I got the same error from initdb. |
#6
| |||||
| |||||
|
|
Hm. It's starting to sound like an actual bug --- though whether it's in HPUX 11.00, gcc, or Postgres is difficult to guess. |
|
The Postgres code is certainly known to work on many platforms, but it could have an unsuspected platform dependency that only shows up here. |
|
One thing I'd ask is whether you are up-to-date on libc patches for your HPUX installation. It's plausible that this could be the result of buggy libc code (such as strtol or related routines). |
|
And my recollection is that 11.00 was about as bulletproof as you'd expect for a dot-zero release :-( |
|
If no joy there, the only thing I can suggest is to rebuild with --enable-debug (if you didn't already) and start digging into it with gdb. If that sounds a bit out of your league, I'd be willing to do it if you can give me ssh access to the machine. (I don't need root, an unprivileged user account is enough.) |
#7
| |||
| |||
|
![]() |
| Thread Tools | |
| Display Modes | |
| |