Network Computingwww.networkcomputing.com

After some work, I had NetSaint functioning--making sure routers were up, mail servers were responding and disk space wasn't filling up. Dozens of plug-ins are available to monitor almost any aspect of any server. I looked at the basics, including ping time, POP3 response, SMTP response and remote disk-space checks. Other plug-ins are available to monitor NNTP (Network News Transfer Protocol), SSH (Secure Shell), Oracle databases, PostgresQL and Hewlett-Packard JetDirect. You can write your own plug-ins for use with the system as well.

Each time I added hosts and services to be monitored, I needed to start up the NetSaint daemon. At that point, the status file of all the hosts and services gets reset. NetSaint schedules everything to be polled, but until everything is polled (sometimes up to an hour later) the elements are in a "pending" state and NetSaint can't warn you of any problems. We would like to see NetSaint accept a "kill­HUP" (hang up) to reread the configuration file and maintain state information. If you need a service monitored immediately, you can use a link on the interface to force a check.

Compile Your Own

Although NetSaint is touted as primarily a Linux package, I had no problem compiling on our Solaris 2.6 systems--installation went smoothly. Then it was time to sift through the configuration, which is a text file listing hosts, services, contacts and commands to perform. Getting the syntax right isn't easy, but a large amount of documentation is available. In addition, I used the mailing list and got answers more quickly than with some technical support hotlines. Setup took a bit longer than that of Ipswitch WhatsUp Gold (which I had also used and was replacing). With the Web administration add-on, called NEAT (NetSaint Easy Administration Tool), configuration is much easier. However, when adding many hosts, I prefer a good old text file.

I added some servers to be monitored, as well as a few routers. For the routers, I wanted to know only if they were running, so a simple ping script was enough to keep track of them. For the servers, I added service-level checks appropriate to each server, such as POP3, SMTP, IMAP or HTTP. As I've done with other node-management software, I entered specific times that the server should be checked and the frequency with which the checks should be done.

Once the configuration was complete, I started up NetSaint and it began to check our servers without complaining. Soon I started to get warnings about our remote sites exceeding the round-trip ping threshold. Because the default script assumes hosts on the local network segment, I had to tweak the configuration a bit to allow for the delays in our diverse WAN. Our next problem came when a secondary mail server went down. It wasn't all that important in the general scheme of things, however, so I simply accepted warning e-mails (sent to a different server) until I fixed the problem.

I did notice that the messages regarding the problem looked a bit odd, though. After checking my configuration, I realized there are supposed to be two different notification formats--one for services (such as POP3 and SMTP) and one for hosts (ping). Once corrected, our messages were more informative.

Because some of our servers require 24x7 uptime, I made sure to add a contact that will page me via e-mail to my cell phone. However, I didn't find NetSaint quite flexible enough for notifications. I was able to configure it to e-mail me only for service problems and e-page me if a server wouldn't respond to a ping. I would like to have service problems escalate. For instance, after NetSaint had warned me a certain number of times via e-mail that IMAP on a server is not responding, it would proceed to e-page me. This feature should be coming in the next release, 0.0.6, according to the company.

Remote Stats

Getting more information from a remote machine is not the same as checking to see if services are running. I wanted to make sure my Unix servers did not run out of disk space, for example. There is a check_disk plug-in, though it only checks local file systems, and there are a few methods of running it on remote Unix machines. Either way, I needed to install the check_disk script on the target machine in order for the process to work.

My first choice was to use SSH or RSH (remote shell) to run the command on the remote machine. I chose to install another NetSaint add-on called NRPE (NetSaint Remote Plug-in Executor). NRPE installs on the target server and lets NetSaint run remote checks. I installed and configured NRPE to allow disk-space checks only from the NetSaint server. To add a bit of security, I configured NRPE to run from inetd (Internet daemon) and added TCP wrappers to the configuration.

Now NetSaint warns me if my root disk gets about 80 percent full and sends a critical message if the disk becomes more than 95 percent full.

Send your comments on this article to Robert J. Kohlhepp at rkohlhepp@nwc.com.