Server Monitoring
Today’s 24 hour service culture requires your servers to consistently perform and your web sites to be ‘always on’. Recognising the added pressure this places on many of our clients, ForLinux have developed a flexible, and yet critical, monitoring service that enables us to support them in achieving their online objectives and meeting their customers’ needs. This service provides real time Linux server reporting to our Technical Services Team, whilst our own Management Servers also use this information to build up a rich historical profile of server performance.
The ForLinux philosophy is not simply to report ’server available’ but to profile common problems. We customise monitoring for your key services, allowing us to make recommendations based on real life service experience.
For example:
- Do you know how well your web server copes at peak traffic times?
- What happens if you then run a mailshot at the same time?
- What is the time it takes to then load the web page?
ForLinux has the answers to these types of questions with a fully customisable, flexible monitoring service designed to not only report errors, but allow better use of your web server resources. All of our technicians have Linux Professional Institute accreditation, giving us the right skills to investigate any problems.
Key Features of the ForLinux monitoring service:
- Flexible – if you have a custom need, we can set up monitoring for it.
- Historical – we build a rolling 12 month alert history so that we can monitor and report upon changes to your web server – if we see something change we will discuss it with you to make sure it is right.
- Manually Reviewed – we don’t just get the alerts, we look at why they happened and discuss any issues we need to with you. Running short of memory all of the time? Is it linked to code issues, or would a memory upgrade resolve some of the email issues you’ve always known about…? We can help to find out.
- Portal Information – some of the core statistics are at your fingertips. All of our clients have access to a comprehensive information portal about their servers and their current and historical performance.
Defining Your Needs
We will work with you to ensure that we are monitoring the things that you want us to monitor. We want to understand your needs and think about what ELSE we can do. It may be that your needs are covered by our standard (but extremely comprehensive) monitoring template, or we might need to sit back and think a little.
Standard Monitoring includes many items, but includes the following key items:
What will the service monitor?
Every minute your server will report the following metrics to our Network Operations Centre.
Server Integrity
- Checksum of /usr/sbin/sshd – This check is performed to ensure the SSH daemon has not been overwritten by a malicious process which would be able to gather usernames and passwords. In most circumstances, hosting providers would remove your server from its network and reformat the server. If the checksum changes an alert is sent. Immediate investigation should be undertaken if this alert is received.
- Free disk space on /home – Free disk space on “/home” is important as it is the normal place to store your web pages. An alert is issued at 95% capacity. This provides an opportunity to login to the server and investigate before becoming critical.
- Free disk space on /var – Free disk space on “/var” is critical to system integrity as it stores log files and generally MySQL. Should the var partition of the server become full the server will not be able to log system events and be unable to store database data. An alert is issued at 95% capacity. This provides an opportunity to login to the server and investigate before becoming critical.
- Free disk space on / – Free disk space on “/” (root) is critical to system integrity. Should the root partition of the server become full the server will become unstable and in extreme circumstances destroy the file system completely. An alert is issued at 95% capacity. This provides an opportunity to login to the server and investigate before becoming critical.
- Free memory – The amount of free memory on your server will have a direct impact on the website performance. Little or no free memory may cause the server to “swap” memory to disk, therefore taking valuable processing time away from your website.
Server and Service Availability
Server and service availability monitors the critical services on your server. The primary services for websites are Web and Database. However, email and FTP can also be critical normal operations.
If any of these services become unavailable an alert is issued. Unavailability is triggered by a local failure. If the network that the server is connected to is down then the “ping to server” is respond.
General server availability is performed by “pinging” the server. This is performed by the server in the Network Operations Centre. If the server in not “pingable” an alert is sent. If a firewall is present, then the ICMP port for the monitoring server IP will need to be allowed.
- FTP server is running
- Email (SMTP) server is running
- web (HTTP) server is running
- Database server is running
- CRON server is running
- Ping to the server (TCP)
- Server Information
Server information is gathered for information purposes only. However, if any of the parameters do change an alert will be sent.
- Host boot time
- Host name
- Host status
- Host uptime (in sec)
- Server Performance
Server performance metrics are taken from several sources to provide an overall system view.
- Incoming traffic on interface eth0 – The amount of network traffic incoming to the server. This is not just web site traffic but the total amount of incoming which includes email, ftp, user logins and other unauthorised traffic.
- Outgoing traffic on interface eth0 – The amount of network traffic incoming to the server. This is not just web site traffic but the total amount of incoming which includes email, ftp, user logins and other unauthorised traffic.
Number of processes – A large number of processes on a server maybe perfectly fine as it may not necessarily affect system performance. However, if a process is “forking” it may disrupt the server and eventually hog the system. An alert is sent once a threshold is breach. This gives the opportunity to login to the server and investigate. - Number of running processes – The difference between the number of running processes and the number of processes is that the running processes are taking more resources than processes not running. Therefore the number of running processes is important to monitor. An alert is sent once a threshold is breach. This gives the opportunity to login to the server and investigate.
- Processor load – Processor load comprise several important system parameters including processor queue, disk I/O, network traffic, memory utilisation and processes. The processor load is a key indicator of performance. Once it reaches a high figure the system can become unresponsive. Therefore it is imperative to be alerted while it is still possible to login to the server.
Call us
Get in touch
Join our mailing list
Training offer
