homespun

libncurses.so.5 undefined symbol _nc_set_no_padding

I was looking to diagnose a high load problem today. For this, I typically use atop. However, I got this output instead...

[root@myserver ~] # atop: symbol lookup error: /lib/libncurses.so.5: undefined symbol: _nc_set_no_padding

I'm running CentOS 6.2, with atop installed from the EPEL repository so I figured it should work. I started looking for another answer. Google is your friend in all such cases, and I found this bugzilla report which implicates zend. Since we are in fact using zend, I thought I would verify the issue.

[root@x-web-01v lib]# which atop
/usr/bin/atop
[root@x-web-01v lib]# readelf -d $(which atop) | fgrep NEEDED
0x00000001 (NEEDED)                     Shared library: [libncurses.so.5]
0x00000001 (NEEDED)                     Shared library: [libm.so.6]
0x00000001 (NEEDED)                     Shared library: [libz.so.1]
0x00000001 (NEEDED)                     Shared library: [libc.so.6]
0x00000001 (NEEDED)                     Shared library: [libtinfo.so.5]
[root@x-web-01v lib]# ldd $(which atop)
        linux-gate.so.1 =>  (0x0016b000)
        libncurses.so.5 => /lib/libncurses.so.5 (0x00b7e000)
        libm.so.6 => /lib/libm.so.6 (0x00e5d000)
        libz.so.1 => /lib/libz.so.1 (0x0072e000)
        libc.so.6 => /lib/libc.so.6 (0x0016c000)
       libtinfo.so.5 => /usr/local/zend/lib/libtinfo.so.5 (0x00110000)
        libdl.so.2 => /lib/libdl.so.2 (0x00e21000)
        /lib/ld-linux.so.2 (0x004f6000)
[root@x-web-01v lib]# nm -D /lib/libncurses.so.5 | fgrep _nc_set_no_padding
         U _nc_set_no_padding

So, sure enough, zend has hijacked libtinfo.so.5. How to fix this - edit /etc/ld.so.conf and add the path to /lib (or /lib64 for 64-bit OS) at the top:

[root@myserver ~]# cat /etc/ld.so.conf
/lib/
/usr/lib
include ld.so.conf.d/*.conf

I added /usr/lib for good measure. Run ldconfig to make the changes live.

Increasing volume under LVM2

Generally, I like to tell a story when I make a post. Today's story is about running out of space on our production mail server. Here is the background. We're running Novell Groupwise on a SLES10, OES2 server serving a small business. There are about 70 mailboxes in the system. We're also highly virtualized, so the server is a virtual machine running as a guest on VMware ESX.

There are two virtual hard drives. The 1st contains the OS, and the 2nd is in a Logical Volume Group. Turns out that was a lucky choice. The 2nd drive was originally provisioned as 110 GB. I knew the drive was getting full, but the last time I checked it had a few GBs free. Yesterday, some urgent system messages were sent to the administrator email address, informing me that the Quick Index couldn't run because the remaining space was 0000 MB. Oops.

HP Smart Array P400 and SLES10 issue

My boss recently dropped an HP ProLiant DL385 G2 on my desk (Literally, he walked in with the thing under one arm!) and asked me to install OES 2 on it. So, I fired up the server and pressed F8 after the Smart Array P400 initialized to configure the array. I set it up as RAID 5 with one logical volume. The server doesn't have a floppy of CD/DVD drive, but I have an external USB CD/DVD, so I put the SLES10 SP1 disk in the drive and rebooted the server. The Installation screen came up and selected Installation, hit Next, etc, etc. All was well for a few moments until I accepted the license agreement. Then the hardware scan came back and reported that there were no hard drives available for installation!

Cutting to the chase - creating the array and logical volume from the SmartStart CD Release 8.10 resolved the issue and the installation completed successfully. Creating the array and logical volumes from the Option ROM utility accessible during boot caused the installation to fail.

The Long version, a litany of things that failed:

• Not what I wanted to see. All the lights on the drives were green, so I shut down the server and reseated the P400 (PCI Express) card and all of the cables. Started up again and failed at the same point. The installer couldn't detect the array.


• Booted up again, deleted and recreated the array. Installation failed.


• Downloaded and installed Smart Array Firmware Maintenance Release 8.20. Install failed.


• Thought it might be the external USB drive, but the CD booted OK, also the HP SmartStart CD booted OK (and recognized the array). There is no SmartStart CD for Linux installations, anyway.


• Downloaded the latest cciss driver for the P400 array card. Copied the files to a USB stick and tried to add the new drivers during Installation. It didn't make a difference.


• Finally, I was browsing the HP Support Forums. There are lots of reports of similar issues when installing RHEL with the P400 smart array card, but one replier mentioned the need to create the array from the ACU (Array Configuration Utility) on the Smart Start CD. I popped in the CD, Release 8.10, deleted and recreated the RAID 5 array, then created a logical volume. Suddenly the Installer could "see" the hard drive and installation was a piece of cake after that.

.NET 3.0 Service Pack 1 adventures

[Listening to: Heartbeats - The Knife - Deep Cuts (03:51)]

I manage a few servers that do not have internet access for security reasons. So, I had documented some time ago how to use Microsoft Baseline Security Analyzer (MBSA) in an offline mode. Recently, I ran into two problems during a security scan.

I had installed .NET 3.0 framework on an isolated server, so when I ran the MBSA security scan recently, it told me I needed .NET 3.0 SP1. No problem I thought. After moving the security scan files back to an internet-connected PC, I clicked on the download link for that file. Well, it directed me to a MSXML6 download instead. That's the 1st time that ever happened! So, I went to the download page for .NET 3.0 SP1. But, there's only a stub executable there. Very nice for minimizing the size of the download to only those components needed by the target machine. Very diasterous for an offline PC!! I needed the full redistributable package, but there was no link to it on that page. The only reference I found on the MS support site was for .NET Framework 3.5, whiich contains the 3.0 SP1. I didn't need that, so I kept looking.

Finally I ran into this article on MSDN blogs. SUCCESS! I downloaded the 3.0 SP1 x86 redistributable.

So, then I went to install it, and got a big disappointment. The installer failed and told me that I needed XPSEPSC installed 1st. So what the heck is that? I found this helpful post which described exactly the same problem. And therein is a link to XPSEPSC.

Installed the XPSEPSC package, followed by SP1 without any further incident.

The Many Faces of Proxy in SuSE Linux SLES

[Listening to: Air Tap! - Erik Mongrain - Fates (03:46)]

Perhaps this will save some other Linux sysadmin some time and frustration...

The first few times I installed SLES9 and OES servers in our environment, I clicked on "Check for Updates" during the installation. Of course the service always failed to connect to Novell's update site, but I didn't know why at first. I soon realized that it was because we were behind a proxy server, preventing the installer from connecting.

Some of the OES servers were configured to use iManager 2.5 or 2.6, but since we still use ConsoleOne for so many management tasks, I didn't realize that the proxy settings for tomcat4 were "broken".

So, I took to configuring the proxy addresses during the installation. Remember, this is SLES9. The updater still couldn't connect because there was no configuration for https:// proxy, which is what the updater actually needs in order to do its work.

Then, SLES9 SP1 (and SP2, etc) came out. The installation program was smarter and now I could put in an https:// proxy setting. However, Novell started to use "rug" to update my OES servers. OK, I had learned how to set the https:// proxy, but rug actually doesn't use those settings. Grrr. So I found a TID that showed how to set the rug proxy.

Eventually, SLES10 comes out and proxy configuration is much more behaved. In fact, it works very well. I installed OES2 as an add-on package for SLES10 recently. Along with that, I installed iManager 2.7. I spent a little more time reading the docs with iman2.7 and so I was pleased to see that it would automatically connect to Novell and download new plugins. Although this was possible in iman 2.6, as I said before, I didn't realize my proxy settings were wrong - actually, they didn't exist. So, here's a variety of proxy settings.

For SLES9/OES:

GUI - use YAST2. Click on Networks Services, then click on Proxy.
Check the box to "Enable Proxy" then fill in your http:// ftp:// and https:// proxy values. Make sure that "localhost" without the quotes is included in the No Proxy Domains box. Click OK. SLES9 before SP1 did not include the https:// proxy setting, if I recall correctly. To configure the SLES9 proxy settings via the command line, using vi, for example:

#vi /etc/sysconfig/proxy

make sure the following lines are present or add them if they're not:
PROXY_ENABLED="yes"
HTTP_PROXY="http://your.proxy.server:8080" (or whatever port is apprpriate)
HTTPS_PROXY="http://your.proxy.server:8080" (or whatever port is apprpriate, add this line if it is not present)
FTP_PROXY="http://your.proxy.server:8080" (or whatever port is apprpriate)
NO_PROXY="localhost"

For rug, at the command-line type:
#rug set-prefs proxy-url "http://your.proxy.server:8080" (or the appropriate port number)

For tomcat4, you have to edit the tomcat4.conf file and add the following lines:

# vi /opt/novell/tomcat4/bin/dtomcat4

CATALINA_OPTS="$CATALINA_OPTS -Dhttp.proxyHost=your.proxy.host"
CATALINA_OPTS="$CATALINA_OPTS -Dhttp.proxyPort=8080"
CATALINA_OPTS="$CATALINA_OPTS -Dhttp.nonProxyHosts=*.mydomain.com|localhost"

and then restart tomcat4 (# rcnovell-tomcat4 restart)

For tomcat5 (iman 2.7), edit this file and add the same three lines:
# vi /etc/sysconfig/j2ee

some references:
Getting the SLES 10 / SLED 10 Software Updater to work if you use a Proxy Server
Configuring Tomcat to use a Proxy server
Novell Doc: OES 1 - Preparing the Server for Patching - actually read section 6.2 for the rug command-line switches.

Updating Suse SLES10 to SP1

[Listening to: Percusien Fa - Erik Mongrain - Fates (03:51)]

Well, I was upgrading a server yesterday from SLES10 to SP1. I started out following the upgrade directions on the Novell site. Since I had the iso images handy, having downloaded them awhile ago, I decided to go with "Update via using and SP1 installation media". It's fairly straightforward, although not exactly as described in the TID (Technical Information Document). I booted off CD1 and followed the prompts for installation. "System Update" doesn't appear till three or four screens into the Installation procedure. Life was pretty painless after that - feeding the CDs as needed.

OK, got the SP1 update installed without a problem - or so I thought! I ran Online Update as well to get all of the latest patches. Rebooted a sufficient number of times - there was a kernel upgrade amongst the other patches, as I recall. Well, then apache2 wouldn't start. Looking in the Apache error log ( /var/log/apache2/error_log ) showed the following error: "Can't load Perl module Compress::Zlib for server webbie:0, exiting" - so I figured I'd fire up cpan and install Compress::Zlib module. But, no. Not that easy. I went into cpan (perl -MCPAN -e shell) and tried to install the module. Instead I got another error: "Use of unitialized value in concatenation (.) or string at /usr/lib/perl5/5.8.8/i586-linux-thread-multi/Scalar/Util.pm line 30."

OK, I really was ticked off, but I was pretty sure that Scalar::Util was actually the broken module - because I ran into a similar problem when upgrading my installation of Request Tracker. Just search for Scalar::Util and Request Tracker and you'll find what I mean within the top few hits. That's what I did and I hit this site. Close enough! I downloaded the tgz file as directed on that page, untarred it, cd'd into the Scalar-List-Utils-1.19 directory, then ran

perl Makefile.PL
make
make test
make install

Finally, started Apache2 without a hitch! Success. /etc/init.d/apache2 start 

w.bloggar portable version

I'm really pleased to share that Marcelo Cabral has released version 4.03 of w.bloggar. For a long time, this has been my favorite blogging tool. Development appeared to stop for about a year, but it looks like things are rolling again. With version 4.03, Marcelo has included a portable version that can be run from a USB stick. Very cool! If you use w.bloggar, you should check it out - and support Marcelo's efforts by making a donation if you can.

Linksys PSUS4 and Vista printing

UPDATE: well, this only worked once. The next time I rebooted, Vista couldn't find the printer again... What a drag. Looks like I won't be using the PSUS4 any longer...

Well, I had purchased a small Linksys printer server (the PSUS4 ver 1.1) awhile ago and it had worked on my old XP Home PC - although it actually didn't support the all-in-one Epson CX6000 completely. Nonetheless, installation had been painless and XP had readily recognized the network printer.

However, I recently bought a new PC that has Vista Home Premium installed on it. Now the fun begins. I plugged my old HP Deskjet 842C into the PC and Vista installed the drivers right away. Then I installed the print server and plugged the printer into it. I installed the Linksys software that came with the device and the setup utility could see both the device and printer. I gave the print server a static IP address rather than going with the default DHCP setting. However, Vista did not automatically detect a new network printer.

It's always a good plan to reboot a Windows PC when you mess with networking - so that's what I did next. After a reboot, Vista still couldn't "see" the networked printer. So, I did what everyone should do at this point... Search Google.

Here's a solution that worked for me.

1. Open the Control Panel in Vista. Under "Hardware and Sound", click on Printer.
2. Click on "Add a Printer."
3. Click on "Add a Local Printer".
4. Under "Use an existing port:", click on an entry that looks like this: "\\Lk38ce86\p1 (Client Side Rendering Provider)"
5. Click Next and choose your printer manufacturer and model - mine was HP DeskJet 840C/841C/842C/843C.
6. Click Next and Use the driver that's already installed - Remember, I directly attached the printer and installed drivers already.
7. Print the test page to convince yourself it's working and then click Close.

Good luck!

w.bloggar rebirths!

Holy crap! There's a new version of w.bloggar! My favorite blogging tool of all time. This is fantastic news. Everyone should support Marcelo Cabral and make a donation if possible! Congratulations, Marcelo! Download the latest version here.

Zoe and Java 6 problem

I've used Zoe for a long time, several years. I think it's a great program. It's no longer being developed, afaik, but it's still very useful at handling my email. Let's see... At the moment, I've got 161,206 messages indexed. Believe it or not, there are a few that go back as far as 1995. Why they will ever be useful, I have no idea... :) Well anyway, ever since I upgraded my Java installation to version 6, part of Zoe has stopped working. At the bottom of the top banner, there used to be some navigation tools but now all I see is this cryptic bit of text -- {v:path}. Since I don't know a thing about java development, I have no idea how to fix it. I do know that if I uninstall the Java 6 and roll back to Java 5, the error goes away and I get the normal navigation cues. The source to Zoe used to be available from the zoe.nu website, and I found a link to it here. So, I guess this is a "lazy web" request. I'd really like someone to grab the source and fix the Java error that keeps Zoe from performing the way it should.