I'm trying to ./compile Perfparse on my Nagios CentOS 4.x box, but am having this error. (And am hoping someone here might point me in the right direction)
"configure: error: required library libgd unusable."
Here is a history of my problems:
First, it ./configure 'd ok, but I couldn't "make && make install" due to an error. I googled and found this FAQ on the the subject (http://perfparse.de/tiki-view_faq.php?faqId=1)
********************************************************* Q: When i try to compile, i get errors like: perfparse-0.105.5/perfparse/storage.c:269: undefined reference to `g_path_get_basename' A: Try this: cd perfparse make copy & paste the "/bin/sh ../libtool ...." line and add "/opt/gnome/lib64/libglib-2.0.so" at the end. The path may vary. Try to find the libglib-2.0.so. Then try again with "make". *********************************************************
Not sure what happened, but it eventually was working, and later I got past that error and got a "gdImage" error when I tried to "make && make install". (This too is on the FAQ page)
********************************************************* "A: The gd library is probably not installed yet on your system. Or you are missing the development package such as libgd-devel or libgd-dev.
Install gd and try again." *********************************************************
Not knowing what to do (gd-2.0.28-5.4E and gd-devel-2.0.28-5.4E were installed), we "yum removed" these packages (which destroyed Nagios, and I had to reinstall and reconfigure that too), and ever since, I've gotten that ""configure: error: required library libgd unusable" error whenever I've tried to ./configure with the automake tools.
Any advice would be greatly appreciated!
On Fri, Nov 30, 2007 at 12:45:47PM -0800, Rogelio alleged:
I'm trying to ./compile Perfparse on my Nagios CentOS 4.x box, but am having this error. (And am hoping someone here might point me in the right direction)
I just downloaded and tried it. Turns our you need the gd-progs package installed to get the gdlib-config script (necessary to satisfy perfparse's configure).
After installing gd-progs, everything built fine.
So verify the following on your machine (make sure they are genuine centos packages and not from a 3rd party repo):
$ rpm -q --qf '%{NAME}-%{VERSION}.%{RELEASE}.%{ARCH}\n' gd gd-devel gd-progs glib glib-devel gd-2.0.28.5.4E.x86_64 gd-devel-2.0.28.5.4E.x86_64 gd-progs-2.0.28.5.4E.x86_64 glib-1.2.10.15.x86_64 glib-1.2.10.15.i386 glib-devel-1.2.10.15.x86_64
Then start over with a clean build and try configure and make again.
On 11/30/07, Garrick Staples garrick@usc.edu wrote:
I just downloaded and tried it. Turns our you need the gd-progs package installed to get the gdlib-config script (necessary to satisfy perfparse's configure).
I have those installed already, but am not sure what is still hosed.
Right now, I cannot get past the configure script. I keep getting the error:
(bunch of OKs, then the following):
checking for gdlib-config... /usr/bin/gdlib-config checking for gdMalloc in -lgd... no configure: error: required library libgd unusable. Install required and dependent files and rerun configure. See config.log or FAQs for details.
Not knowing what else to do, I reinstalled (via rpm -i) the following packages
gd-2.0.28-5.4E.i386.rpm gdk-pixbuf-0.22.0-17.el4.3.i386.rpm gdbm-1.8.0-24.i386.rpm gd-progs-2.0.28-5.4E.i386.rpm gdbm-devel-1.8.0-24.i386.rpm gtk+-1.2.10-36.i386.rpm gd-devel-2.0.28-5.4E.i386.rpm php-gd-4.3.9-3.22.4.i386.rpm
Here is what I got when I tried to "rpm -i" install the gdk-pixbuf
rpm -i gdk-pixbuf-0.22.0-17.el4.3.i386.rpm error: Failed dependencies: libgdk-1.2.so.0 is needed by gdk-pixbuf-0.22.0-17.el4.3.i386 libgtk-1.2.so.0 is needed by gdk-pixbuf-0.22.0-17.el4.3.i386 Suggested resolutions: /home/buildcentos/CENTOS/en/4.0/i386/CentOS/RPMS/gtk+-1.2.10-36.i386.rpm
I then installed that gtk rpm, and I got the following issue:
rpm -i gtk+-1.2.10-36.i386.rpm error: Failed dependencies: gdk-pixbuf >= 0.18.0 is needed by gtk+-1.2.10-36.i386 Suggested resolutions: /home/buildcentos/CENTOS/en/4.0/i386/CentOS/RPMS/gdk-pixbuf-0.22.0-17.el4.3.i386.rpm
So, I ran "yum install gdk-pixbuf", which then grabbed the gdk-pixbuf and gtk+ packages.
Now, the only gd package that has a problem is php-gd (which I'm not even sure is related)
rpm -i php-gd-4.3.9-3.22.4.i386.rpm error: Failed dependencies: php = 4.3.9-3.22.4 is needed by php-gd-4.3.9-3.22.4.i386 Suggested resolutions: /home/buildcentos/CENTOS/en/4.0/i386/CentOS/RPMS/php-4.3.9-3.22.4.i386.rpm
Should I perhaps be looking in the direction of ldconfig to fix my problems?
On Sat, Dec 01, 2007 at 09:13:50AM -0800, Rogelio alleged:
On 11/30/07, Garrick Staples garrick@usc.edu wrote:
I just downloaded and tried it. Turns our you need the gd-progs package installed to get the gdlib-config script (necessary to satisfy perfparse's configure).
I have those installed already, but am not sure what is still hosed.
Right now, I cannot get past the configure script. I keep getting the error:
(bunch of OKs, then the following):
checking for gdlib-config... /usr/bin/gdlib-config checking for gdMalloc in -lgd... no configure: error: required library libgd unusable. Install required and dependent files and rerun configure. See config.log or FAQs for details.
So what does config.log have about this error?
Not knowing what else to do, I reinstalled (via rpm -i) the following packages
gd-2.0.28-5.4E.i386.rpm gdk-pixbuf-0.22.0-17.el4.3.i386.rpm gdbm-1.8.0-24.i386.rpm gd-progs-2.0.28-5.4E.i386.rpm gdbm-devel-1.8.0-24.i386.rpm gtk+-1.2.10-36.i386.rpm gd-devel-2.0.28-5.4E.i386.rpm php-gd-4.3.9-3.22.4.i386.rpm
Aren't you on an x86_64 machine? Why are all these packages i386?
Reinstalling packages never fixes anything (especially random, unrelated packages like gdbm and gtk). Just use yum to get the packages again and put your system back together again.
On 12/1/07, Garrick Staples garrick@usc.edu wrote:
Aren't you on an x86_64 machine? Why are all these packages i386?
Reinstalling packages never fixes anything (especially random, unrelated packages like gdbm and gtk). Just use yum to get the packages again and put your system back together again.
Someone else installed it (in a datacenter far away), and I don't believe that they installed the 64 bit ver of CentOS. Do I need to change something before this script will run?
On Sat, Dec 01, 2007 at 11:27:56AM -0800, Rogelio alleged:
On 12/1/07, Garrick Staples garrick@usc.edu wrote:
Aren't you on an x86_64 machine? Why are all these packages i386?
Reinstalling packages never fixes anything (especially random, unrelated packages like gdbm and gtk). Just use yum to get the packages again and put your system back together again.
Someone else installed it (in a datacenter far away), and I don't believe that they installed the 64 bit ver of CentOS. Do I need to change something before this script will run?
Your original email had a "lib64" path, implying that had a 64bit machine, which if you were manually installig 32bit packages would explain everything.
'uname -m' will tell you if have 32bit or 64bit OS. Don't guess.
No, nothing special needs to be done other than having the correct devel packages installed. I suspect you've got things mucked up at this point.
On 12/1/07, Garrick Staples garrick@usc.edu wrote:
Your original email had a "lib64" path, implying that had a 64bit machine, which if you were manually installig 32bit packages would explain everything.
I'm sorry for the confusion. My original email was quoting the FAQ on Perfparse's documentation, not reflecting my box's output
'uname -m' will tell you if have 32bit or 64bit OS. Don't guess.
uname -m -> i686 (as does /proc/cpuinfo).
No, nothing special needs to be done other than having the correct devel packages installed. I suspect you've got things mucked up at this point.
I'm sorry, I'm confused. When I run "rpm -q", I get all i386 packages (for everything on that box). The 64 bit stuff was not* installed.
The problems started occuring when another admin "yum remove" 'd gd and gd-devel, which also removed all Nagios related stuff. I yum installed Nagios, and when I went to configure, I got this libgd unuseable error.
Before this snafu, my problem was related to make, not related to configure. Now I've gotta fix this problem with configure before I can go back and figure out what my problem was before.
As far as I know, this is a corrupt package issue, and I'm hoping that there is fairly non-disrputive way of reinstalling everything required for PerfParse.
On Sat, Dec 01, 2007 at 02:04:29PM -0800, Rogelio alleged:
On 12/1/07, Garrick Staples garrick@usc.edu wrote:
Your original email had a "lib64" path, implying that had a 64bit machine, which if you were manually installig 32bit packages would explain everything.
I'm sorry for the confusion. My original email was quoting the FAQ on Perfparse's documentation, not reflecting my box's output
'uname -m' will tell you if have 32bit or 64bit OS. Don't guess.
uname -m -> i686 (as does /proc/cpuinfo).
No, nothing special needs to be done other than having the correct devel packages installed. I suspect you've got things mucked up at this point.
I'm sorry, I'm confused. When I run "rpm -q", I get all i386 packages (for everything on that box). The 64 bit stuff was not* installed.
Ok, no more 32bit vs. 64bit confusion on my part.
The problems started occuring when another admin "yum remove" 'd gd and gd-devel, which also removed all Nagios related stuff. I yum installed Nagios, and when I went to configure, I got this libgd unuseable error.
Before this snafu, my problem was related to make, not related to configure. Now I've gotta fix this problem with configure before I can go back and figure out what my problem was before.
As far as I know, this is a corrupt package issue, and I'm hoping that there is fairly non-disrputive way of reinstalling everything required for PerfParse.
config.log has the actual error that will tell us the problem.
On 12/1/07, Garrick Staples garrick@usc.edu wrote:
On Sat, Dec 01, 2007 at 02:04:29PM -0800, Rogelio alleged:
The problems started occuring when another admin "yum remove" 'd gd and gd-devel, which also removed all Nagios related stuff. I yum installed Nagios, and when I went to configure, I got this libgd unuseable error.
Before this snafu, my problem was related to make, not related to configure. Now I've gotta fix this problem with configure before I can go back and figure out what my problem was before.
As far as I know, this is a corrupt package issue, and I'm hoping that there is fairly non-disrputive way of reinstalling everything required for PerfParse.
config.log has the actual error that will tell us the problem.
After talking my problem over with several other CentOS ppl, I think that my best solution is to just either (a) compile GD from scratch, or (b) scrape box and redo (since it's not in production yet).
Anyone have any suggestions for (a)? Fortunately, I have my NagiosQL mysql dump, so restoring all of my Nagios files will be really easy once I rebuild it (which I can do in like 2 hours now).
On Sun, Dec 02, 2007 at 03:04:36PM -0800, Rogelio alleged:
On 12/1/07, Garrick Staples garrick@usc.edu wrote:
On Sat, Dec 01, 2007 at 02:04:29PM -0800, Rogelio alleged:
The problems started occuring when another admin "yum remove" 'd gd and gd-devel, which also removed all Nagios related stuff. I yum installed Nagios, and when I went to configure, I got this libgd unuseable error.
Before this snafu, my problem was related to make, not related to configure. Now I've gotta fix this problem with configure before I can go back and figure out what my problem was before.
As far as I know, this is a corrupt package issue, and I'm hoping that there is fairly non-disrputive way of reinstalling everything required for PerfParse.
config.log has the actual error that will tell us the problem.
After talking my problem over with several other CentOS ppl, I think that my best solution is to just either (a) compile GD from scratch, or (b) scrape box and redo (since it's not in production yet).
Anyone have any suggestions for (a)? Fortunately, I have my NagiosQL mysql dump, so restoring all of my Nagios files will be really easy once I rebuild it (which I can do in like 2 hours now).
Rather than look at config.log? Have fun with that.
On 12/2/07, Garrick Staples garrick@usc.edu wrote:
Rather than look at config.log? Have fun with that.
I looked at the config.log and didn't really know what I was looking at since I'm fairly new to autotool-related errors. Since this has to be in production fairly soon, I'm thinking of cutting my losses or perhaps rebuilding from sourdce (I started another thread on the pros and cons of that approach).
I showed this to other CentOS admins, and they suggested I cut my losses and quickly rebuild.
Thanks for your help, Garrick!
On Sun, Dec 02, 2007 at 10:20:56PM -0800, Rogelio alleged:
On 12/2/07, Garrick Staples garrick@usc.edu wrote:
Rather than look at config.log? Have fun with that.
I looked at the config.log and didn't really know what I was looking at since I'm fairly new to autotool-related errors. Since this has to be in production fairly soon, I'm thinking of cutting my losses or perhaps rebuilding from sourdce (I started another thread on the pros and cons of that approach).
I showed this to other CentOS admins, and they suggested I cut my losses and quickly rebuild.
Thanks for your help, Garrick!
Just send me the config.log offlist.
-
Garrick Staples -
Rogelio