5.3 Server configuration

The page “03 Server configuration” is available only for the administrator users, here the user can add new, delete and modify servers. Adding and deleting a new server is possible for measurement, SOAP, secondary application manager, secondary event receiver, additional WEB (FLASHJS) and additional chart servers. Adding and deleting of tenant threshold and report servers is possible, but these configurations must be done on the tenant configuration page.

The table on the starting page contains the following columns:

·       Editing links:

o   [add new]: adding a new server

o   [edit]: editing a server

o   [del]: deleting a server

·       ID: unique identifier of the server. It is used when starting and halting the server from the command line, or when displaying the summary of the collected data for the measurement servers.

·       Name: the name of the server

·       Status: status of the server: running, not running, or unknown; see next subsection for details

·       Type: the type of the server; or the type of the server group for collector servers

·       Server group: for a measurement server the name of its group

·       Host: the host computer of the server

·       Login: the user name used to login to the given machine

·       Password: the password of the user. If it is not set, the application can still log in if SSH key-based authentication is configured

·       Directory: the directory of the PVSR application. The name of the directory should be an absolute path (i.e. it should started with /)

·       Param1 and Param2: Further parameters whose meaning depends on the type of the server (they are displayed with the appropriate name):

o   Measurement server:

§  Persistent SSH and SFTP connection (Param1): Whether the SQLLDR and the MANAGER server can keep a persistent SSH and SFTP connection with the measurement server or not. Persistent connection is strongly recommended since superfluous logins and logouts can be avoided. There is only one case when non-persistent connection is necessary: in the case of certain former SSH implementations the established SFTP connection has fully utilized the CPU of the target machine, even when no operation was executed

§  The number of additional seconds to be waited for the results if the collector does not run (Param2): If the measurement server is not running then the central data collector always waits a measuring cycle for the results, if there is a server that is running (e.g. if there is a server that is not running and a measurement of one-hour cycle belongs to it, and the current time is 11:59, then the data collector server accepts the measurement made at 11:00). This waiting period can be increased on a server-by-server basis (of course this causes further delay in threshold and report calculations since those have to wait for all measurements to arrive)

o   Application manager:

§  Cycle (in sec): The number of seconds waited between server monitorings

§  Number of files: During self-testing it checks the number of files in the tmp/done directory of the PVSR that executes the data loading component. If this is bigger than the value specified here, then the system generates a server event and notifies the operator.

o   Data compression and moving and partition creation: the time of running (HH:MM): These servers execute operations once a day (though they are continuously running) – the time of their operation (partition creation is an exception since the daily operation is executed immediately at startup)

o   Threshold processing and report creation: the maximum time period (in hours) of former data to be processed at startup. At startup both of these servers determine the time up to which the individual thresholds as well as report variables were processed, and if the hour of the last processing is smaller than the current hour minus the specified time period, then they consider the processing to be done up to the current hour minus the specified time period. Example: if the five-minute thresholds were processed by the server up to 8:55, and it is restarted at 18:05 (same day), and the set value is 6, then processing will only be continued from 12:00

• The Threshold processing modules have an additional parameter: the protocol and port pair, where the irregular data collectors and the collectors using the quick threshold evaluation feature send the measured values to be used in the threshold evaluation. The value of the parameter is either just a number in which case the protocol is TCP, or in any of the format: udp/<port>, tcp/<port> or ssl/<port>. If the SSL protocol is used then additional configuration must be done, see the QUICK_EVAL_THRESHOLD_SSL… parameters in CONFIG_INI.pm

o   SQLLDR: the number of hours the data loading cycle should finish in. See section 5.5

o   Backup: the first parameter is the backup role, the second is the address of the remote (primary or backup) PVSR

During the configuration of the servers the following fields can be modified: name, host, login, password, directory, enabled/disabled state and the optional Param1 and Param2. Out of the parameters the host-login-directory triad can only be modified if the given server is not running. If a server is disabled then the system stops it if it is running. Disabled servers are not even listed on the status and watchdog pages.

IMPORTANT: although each server module can be disabled, it is highly recommended that non data collector servers should only be disabled temporary, since if such a server is not running then it will have serious effects on the system!

The parameter list on the modification page is completed with the server group as well when a measurement server is added.