382295 Server monitoring/availability

Comprehensive system to measure the availability of the server, hosting... I need an application to measure the availability of web services. It should be a fully automated system with a large range of settings for paying and free users (depending on the settings). I need a safe, fast, reliable and well-arranged application that will not place many requirements on the server, that can be set easily and in great detail and will be operated on servers of its own. • option to measure in an interval of 10 minutes to 15 seconds • measurements from diverse servers: proxy (different countries) + a function making sure that the proxy is working • notification about failures by email, as a XML or RSS feed This notice will be sent 1x in case of a longer-lasting uninterrupted error. It means that if the failure takes a half-hour the customer does not receive 30 emails with error messages in the email box but only one message indicating the time of the start of the failure and some further information. • option to set a maximum response time • professional statistics • applets of their own for browsers including statistics • detailed data about each and every failure • display of the server response time • well-arranged availability graphs • consolidation of several failures into one • many options of setting the failure notification • daily/weekly/monthly spastics by email • option of public statistics • export of availability in icons, XML or plain text • discounts for measuring several services • measurement of services including the sign-on (FTP, POP3, MySQL...), option to set the PORT, option to set the timeout 1 - 30 seconds HTTP (head) HTTP (get) HTTPS (get) FTP FTP-SSL POP3 IMAP IMAPs SMTP SMTP (+EHLO) SMTPs SMTPs (+EHLO) MySQL PostgreSQL Ping LDAP LDAPs DNS server IPv6 For every address it is possible to receive notifications about failures or daily, weekly or monthly statistics. Option to have repeated notifications in case of a continuing failure or a message with a data summary after the failure has been corrected. Display of statistics Record the rate of failures/measurements in the selected time interval, returned HTTP codes, consolidate several failures into one and sort everything into hourl.y/dail.y/monthl.y periods of time. This should all be displayed in well-arranged statistics of availability (administration), all statistics can (but need not) be viewed publically. • Catalogue of web sites that the owner did not select as private • Moreover the measurement results can be exported through the RSS feed, an XML file or it is possible to apply the main availability number from the text file or display an availability icon. • Should the measurement server be disconnected from the Internet, no measurement takes place. ------------------------------------------------------------ Absolute availability is a mathematically exact figure - there is a number of failures (during a certain period of time) and there is a number of measurements. If we divide these two values, we get the value of availability within the period of measurement. Common availability is based on the absolute availability but takes into account several aspects and assigns them a different level of importance: • When during the year was the value measured: failures (and the data as such) are all the less important the earlier they were obtained. The data obtained today have the significance 1, the data obtained 3 months ago are only half as significant (0.5), the data from half a year ago are four times less significant (0.25) and the data obtained 9 months ago have no significance. Basically what happened this month is more significant than what happened about a year ago, the course is fluent and linear. (option to set in the administration) • Availability at night: data obtained between 2 am and 6 am (option to change for every address) are half as significant as other data. The assumption is that considerably fewer users are surfing the web on the European Internet at this time, which means that potential unavailability does not matter that much. • Option to set the time of archiving the data ------------------------------------------------------------ • All users registered in the system have their own credit, i.e. an amount of money that they paid and can use. Every minute a proportional share with regard to the current amount of monthly payment is deducted. For example if the user pays 50 per month, approximately 0.0069 is deducted every minute. This system guarantees that if the service is used for only 3 days, 2 hours and 23 minutes, the user really only pays the price for measuring 3 days, 2 hours and 23 minutes, not more. Moreover, it is possible to change the range of measurements as you like and continuously add/cancel/change services without any problem or financial loss. Generally every minute a different amount will be deducted, you pay no extra money. The credit and the current price for measurement can be viewed in the administration. No difference is made between what method is used to measure but a difference is made between the number of measurements made. In case of measuring several web pages, there will be discounts that I can set in the administration just like the price of the services. Example: 1 measurement = $1, 10 measurements = $8 (discount 20%)……. • If a failure is detected and measured, the users should have a chance to run a user script. This way the customers will be able to automatically activate some services. • The condition for it is multilingualism; every language will be operated at its own domain (one system + a separate domain for every language + currency) • Recording the path status to the measured server at the moment of failure. • Option to change the graph colors. • Marking the individual failures with a text. • Display of graphs in any time range chosen by the user • Inform about little credit • Option to set the user rights for various groups of registered users Example: Option to set only some services, not all services, for free users • Option to check the key word from the web - ability to test the contents of the returned page for the specified string • Invoice in PDF • Payments through Paypal, bank transfer or Moneybookers • Option to set the time of sending the email with a summary for the whole day • UTF8 coding • Icon if the server is available or not • Option to send statistics and more information to several email boxes that will be separated by comma • Option to set a different mode of sending the failure summary starting at a minimum number of failures (or a minimum failure duration). I use alarming text messages but as the 1st failure means nothing I want to ignore it. With the current settings, even if no text message is sent, a summary of its start is sent every time. Therefore, I would like to propose a change meaning that the summary could not be sent before at least the 2nd failure, etc. • Every user receives an email after registering with his/her data Must support PHP5.2.x In case of free, it is possible to use the service without any time limitations and for up to 30 web pages (according to the settings in the administration). How the measurement is done: The server tries to connect to the measured server at a specified moment. If it succeeds, everything is ok, the measurement is recorded as available or as an error code (depending on what the server returns). If the measurement expires in timeout or if it is rejected by some element on the way, our server gets connected to another server which is physically located somewhere else (proxy) in order to try and connect to the measured server. If even this is not possible, we check the connection to the Internet - by calling several sites that should always be on-line. At least one of the servers should respond at every location. And only then is the measurement marked as a failure. If some part of the measuring mechanism fails, the measurement is not recorded in the system at all (no availability and no failure are recorded). This is our protection against false failures. Everything must be tested and immediately ready to use in real operation.