by Krystian Rybka 
July 09. 2018 No matter of period of time (or the universe - remember Varys from Game of Thrones?) knowledge is always a valuable good. Knowledge consists of the great amount of information. Have you ever wondered on the name "Information Technology"? It means that owning and processing data is a base of the running of systems. Today we are focusing on monitoring of servers. Owning a nice garden is only the half of the success. To keep this state, you need to be aware of what is going on in every corner of it, the ability to predict potential disasters or dangers. And you better have great tools for that.
Server monitoring is a wide subject, that is why I decided to narrow it to a few tools which do their job in my everyday work. The tools can be divided according to models. In this case, it is going to be SaaS (web service) and On Permise (traditional). These solutions are fully or partially free. Every one of them works basing on agent - small application installed on a monitored server which sends data to the main monitoring server.
Huge advantages of Software-as-a-service monitoring are the simplicity of configuration, independence from its own infrastructure, scalability and flexibility for changes in the infrastructure. However, there is also a dark side: the cost of this solution. For example, New Relic - leading-edge provider of monitoring services - charges 100$ for a month of thorough monitoring of a single instance size A1 Standard in Microsoft Azure. Is it expensive? Yup. But wait, you need a constant monitoring? Definitely. Without advanced options offered by New Relic? Guess so. Without paying anything? HELL YES! Well, I have good news.
Don't get discouraged by the fact that it is a beta version. Pinguzo from Softaculous team is a great solution as an essential tool for monitoring of your basic servers. The configuration is easy - register, add a new server, fill in some information, run a displayed command and... oh wait, there is no other step. The command installs Pinguzo agent which downloads data about the state of the server. After a minute Pinguzo dashboard will display data about CPU, RAM, running processes, free HDD space or the number of active connections to the server.
In case of crossing given values, Pinguzo will send an email to the user or group of users.
An additional functionality is the monitoring of running websites - Pinguzo checks websites chosen by you. If the HTTP server returns a status different than "200 - OK", Pinguzo will inform you about the anomaly.
In a tab called "Checks", you can monitor postal services, SSH, FTP, RDP, MYSQL base or a port of your choice.
Okay, time to stop this sugariness. There are a few disadvantages:
- Pinguzo supports only Linux systems.
- notification emails for users may arrive behind schedule, which may vary - it can be both a minute and several dozens of minutes. You have to keep in mind that Pinguzo’s mail servers have their limits.
- from time to time fake alerts appear - Pinguzo still runs as a beta version so some service issues can happen. This is the moment when we get notifications that our server is unavailable, though everything is fine with it.
Apart from these faults, Pinguzo still has a strong advantage which is being completely free. You definitely should give it a shot.
Monitoring of Windows system is something Pinguzo cannot do (yet). So, how to deal with taking care of server working under control of the system from Redmond?
Anturis is also a SaaS solution. It deals with both Linux and Windows. Unfortunately, it is paid. However, Anturis allows for using five probes for free. I use it because of function absent in Pinguzo - monitoring internal system processes. Let's suppose you own a machine running 24/7 on Windows server with some .exe process and you want to be notified about its crash immediately. All you need to do is to install the agent of Anturis (every admin can do this in a blink of an eye), add a component and determinate when the alarm should be brought. It is worth mentioning that Anturis supports the service of JVM environments and allows for monitoring Java applications easily.
A tool that does great on our servers is Zabbix. It is an enterprise-class mature monitoring tool and it’s for free! Need more cases? Okay, let Zabbix show what it gots. Stability thanks to LTS version, active community and development of Zabbix, many methods of monitoring, web interface and data visualisation, access to API and scalability (e.g. adding proxy servers gathering data in order to lighten main monitoring server load).
Before you begin let me tell you about two things:
- you want to monitor up to 500 hosts? Great, you only have to have your own server with web stack (e.g. LAMP) with at least 2 CPU and 2GB of RAM.
- Zabbix is not as easy to configure as SaaS solutions - you probably will have to spend several hours on its installation and both getting to know the interface and its possibilities. But no worries! The Internet has your back with tons of guides that will make the whole introduction easier.
Zabbix runs basing on the agent-server connection. The agent is flexible enough to allow for creating your own methods and "checks" and finally sending them to the monitoring server. You can find lots of pre-made schemes created by Zabbix community. The difficulty level of installation is various because every solution is individual. No worries, again, usually some installation manual is attached to the schema. In theory, you can create an alert for everything - all you need to know is how to script it.
The elasticity of Zabbix allows it for a cooperation with tools like Grafana.
"The tool for beautiful monitoring and metric analytics & dashboards" - the official description of Grafany explains what are we dealing with. One peek at the picture below will resolve all your doubts.
It is also an open-source tool, downloading data from the source which can be an InfluxDB, Elasticsearch or Zabbix. Wait, what?! Zabbix?! Yup. On the Internet, you can find many guides explaining the connection of Zabbix and Grafana. Long story short: in Zabbix, you create a new user with rights of reading the data, then in Grafana, you activate the plugin called "Zabbix", fill in basic information about the connection, make sure that hosts can communicate by 80/443 ports and expect success. The next step is creating your own dashboard in order to display data in form of charts.
Finally, it is time for Cybertec's solution, made for monitoring of PostgreSQL databases using Grafana. I recommend them because of quick deploy (thanks to Docker) and a possibility to monitor the most important values of these databases. The requirement is adding the user of monitoring (which is serviced by the pgWatch2). Adding extension called "pg_stat_statement" is optional, but it allows for a deeper insight into the running instance of the database. The whole is well described in the official guide made by Cybertec. You can do the installation on your own or simply use the ready Docker pack. You can download it here: https://github.com/cybertec-postgresql/pgwatch2.
All of these network monitoring tools have been tested in fights against errors. They work out as the protection of our servers - you should give them a chance and experiment! The solutions will repay you with the peace of mind and the awareness that something takes well care of your infrastructure.
What Is Consul? - The Simple Explanation You Desire
Have an idea ? Let’s talk
Fill up the form and we will contact you shortly.
ServoCode Sp. z o.o.
Jasionka 954E, 36-002 Jasionka, Poland
NIP: 8133719852
REGON: 364182909
KRS: 0000611643
Your message has been sent!
Your message has been sent!