Database monitoringDatabase monitoring - DB check for access and regular execution of the specified query.
Database monitoring feature allows to run a query during every check and react in the appropriate way on the result. Also, it is possible to check just possibility to connect to DB - by ignoring the query field. To set the monitoring, fill the connection data: DB server address, port, database name, login and password of a user for connection. We strongly recommend to create the new user with limited rights. However, do not forget to provide him enough rights for performing the supposed actions. Also, it is necessary to add HostTracker servers addresses, to the whitelists on firewall or other blocking software, to allow the access. The addresses are permanent, and are listed on the same form.
There could be arbitrary query - SELECT, UPDATE, DELETE, INSERT, execution of stored procedures (like scheduler), results comparison, logical operations. The only restriction is execution time - it should not take longer than 30 seconds. Otherwise, the timeout error will be reported.
It is recommended to create the queries which display the necessary value in the first row of the first column. This result can be analysed. For UPDATE, INSERT, DELETE queries the number of affected rows is analysed. There are different ways to analyse the resulted value by comparison with specific preset - equal/not equal/higher/lower/in range. In case the condition is not satisfied, no connection to DB, query timeout - the error is reported.
As many people know, HostTracker is a sites efficiency monitoring system. One of its main functions is to notify the user of any problems promptly. The efficiency of the notifications and the acceptable level of “detalization” are important. If you send alerts at each “sneeze”, the person will not find the important information in this flow...
I was woken up by an SMS at three a.m.
My site dropped for three minutes, and it raised back itself.
But I could not go back to sleep. True-life story
As many people know, HostTracker is a sites efficiency monitoring system. One of its main functions is to notify the user of any problems promptly. The efficiency of the notifications and the acceptable level of “detalization” are important. If you send alerts at each “sneeze”, the person will not find the important information in this flow.
We have provided several mechanisms that will help the right people to get the necessary notifications:
Separation of the notifications into several groups according to their criticality;
No notifications at short-term failures;
Report the problem to the manager promptly;
Report a prolonged failure to the administration;
Use the free alerts first – email, gtalk, and then the paid ones – SMS or phone call;
At the contact level – set the working time when this contact should receive the alerts.
There are three types of notifications:
The website has “dropped”;
The website is still “down”;
The website “rose
The “dropped” and “rose” are clear. The notifications “site is still down” are sent at each test fail, but only at the confirmed drops. The fails confirmation algorithm was described in the article “False alerts exclusion”
For each site-contact pair you may enable or disable the appropriate notification type. The setting can be located in the contact properties as well as in the general “matrix” at the “Notifications subscribtion” page.
Escalation and the notifications detalization level.
Suppose, two people are responsible for the site:
Let's try to implement the following scenario:
In the event of a “drop” we want to send an email message to the administrator immediately;
If the site does not rise within 15 minutes, we send an SMS to the administrator;
If the site is “down” for more than an hour, then we send an SMS to the manager.
Adding the contacts for the users. While adding, draw attention to the “Notification Delay” window.
We appear to have three contacts with the following delays:
Administrator (email) – no delay;
Administrator (SMS) – 15 minutes delay;
Manager (SMS) – 1 hour delay.
According to this configuration the administrator will get all the failures notifications to the email, but SMS notifications will be sent only if the site is “down” for more then 15 minutes. The manager will receive only SMS about major failures lasting more than an hour. Setting up the contact working schedule
Suppose that one administrator can not cope, and we hired one more administrator. The first one works during the first half of the week, the second one works during the second half. Accordingly the notifications should be sent to the administrator “on duty” To set this scenario the window “Set the contact working hours” is used in the contact settings.
In this case the first administrator will receive the SMS notifications from Monday to Thursday inclusive. Additionally, you may divide the notification for different employees according to the time of day, for example appointing day and night administrators.
Conclusions: with the help of relatively simple mechanisms we may cover most notifications fine-tune user scenarios.