Thanks for the suggestions.

I will make sure I change the --with-user and --with-group options before 
this goes into production.

On the --with-db=gdbm issue, your also right. Setting --with-db=text gets 
around this issue completely and this is probably the way I will go.

Just for kicks though I also tried a compile on a RedHat 7.2 box with the 
following options:

./configure --prefix=/usr/local/amanda --libexecdir=/usr/local/amanda/lib 
--with-configdir=/proj/amanda --with-suffixes --with-fqdn 
--with-bsd-security --with-config=NONE --with-tape-device=/dev/rmt/0mn 
--with-smbclient=/usr/bin/smbclient --with-user=root --with-group=root 
--with-db=gdbm --with-debugging --with-includes=/usr/local/include 
/usr/include/gdbm --with-libraries=/usr/local/lib

and got the same result:

----------------------------------------------------------
gcc -DHAVE_CONFIG_H -I. -I. -I../config -I../common-src -I../tape-src 
-D_FILE_OFFSET_BITS=64 -I/usr/local/include -I/usr/include/gdbm 
-D_FILE_OFFSET_BITS=64 -g -O2 -c infofile.c -o infofile.o
infofile.c:46: parse error before `*'
infofile.c:46: warning: data definition has no type or storage class
infofile.c: In function `open_infofile':
infofile.c:391: warning: assignment makes pointer from integer without a 
cast
infofile.c: In function `get_info':
infofile.c:525: incompatible types in assignment
infofile.c: In function `get_firstkey':
infofile.c:554: incompatible types in assignment
infofile.c: In function `get_nextkey':
infofile.c:597: incompatible types in assignment
infofile.c: In function `put_info':
infofile.c:658: `DBM_REPLACE' undeclared (first use in this function)
infofile.c:658: (Each undeclared identifier is reported only once
infofile.c:658: for each function it appears in.)
make[1]: *** [infofile.lo] Error 1
make[1]: Leaving directory 
`/home/mpdavig/stuff/downloads/amanda-2.4.2p2/server-src'
make: *** [all-recursive] Error 1 
----------------------------------------------------------

Is "gdbm" support just broken in this release? Or is there something wrong 
with gdbm-1.8.0 itself?

Has anyone found a solution to this problem?

-- 
Thanks,
Marc Davignon
Systems Administrator, Merrill Lynch ITG

On Fri, 12 Apr 2002, John R. Jackson wrote:

> >I'm compiling with the following params:
> >...
> >--with-user=root --with-group=root
> 
> Side issue.  Do you really need to run as user root?  Amanda is not
> normally set up that way (why ask for permissions you don't need).
> 
> >--with-db=gdbm
> 
> Sigh.  At 2.4 a new database format, "text", was introduced.  It's **so**
> much easier to work with, it was made the default and I doubt anyone has
> used "real" DBM files for years.  There never was enough load on those
> files to justify a database engine (IMO).
> 
> So, rather than digging into making DBM work again (and as I recall, it's
> going to be completely removed before too long), how about switching to
> --with-db=text (or just leaving it off)?  All you have to do (as I recall)
> is run "amadmin <config> export" on your old config before upgrading,
> then "amadmin <config> import" afterward.
> 
> Ummmm, maybe you shouldn't have nuked the old 2.3.0.4 version yet?  :-)
> 
> If you really, really, have to have DBM support, it can probably be
> fixed up.  Although I'll be away at a conference next week (at the UofMD,
> birthplace of Amanda, I just realized :-) and won't be able to help
> until after that.
> 
> >Marc Davignon
> 
> John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
> 

Reply via email to