Monitoring

This section describes how you can monitor your Exasol 8 system through system log messages.

Logs in Exasol 8

Exasol 8 does not have an internal syslog service. Messages generated by the components in Exasol 8 are instead forwarded to the host syslog service. To monitor your Exasol 8 system, you can use a third-party syslog management/monitoring tool to subscribe to the relevant syslog messages.

For more information about how to set up logging, refer to the documentation for the respective syslog tool.

Syslog message structure

The syslog messages generated by the Exasol services have the following structure:

syslog message structure

Source Description
ConfD Events in the ConfD administration interface.
Health

Events related to the performance of the host system. This does not include storage devices.

Suggested values to monitor are: load (CPU load), mem (RAM utilization), and swap (swap utilization).

Authentication Attempts to authenticate to the host.
DWAd Events related to a specified database, such as starting and stopping the database.
Example:
May 30 08:03:18 ip-10-0-5-202 Authentication: 11: [confd] Successful authentication with username root
May 30 08:03:18 ip-10-0-5-202 Health: 11: Load 0.42431640625:0.83349609375:0.71142578125; freq: 2500.0, max 3202.069; cpus: 2; mem: 13.7159 GiB of 15.3379 GiB free; swap: 10 GiB of 10 GiB free; hp: 0 B of 0 B free, 0 B rsvd, 0 B anon; cache: 445.8359 MiB
...
May 30 08:05:01 ip-10-0-5-202 ConfD: 11: n0011 User root submitted db_list (11.111)
May 30 08:05:01 ip-10-0-5-202 Health: 11: Got the db list: ['Exasol']
...
May 30 08:07:27 ip-10-0-5-202 DWAd: 11: [Exasol] User 0 requests new system setup.
May 30 08:07:27 ip-10-0-5-202 DWAd: 11: [Exasol] User 0 requests startup of system.
May 30 08:07:28 ip-10-0-5-202 DWAd: 11: [Exasol] System started successfully in partition 75.

Thresholds

Since Exasol 8 does not have an internal syslog service, you have to set the desired warning thresholds for the monitored values in your syslog management tool. The actual threshold values to use depends on your system configuration.

For example, a good starting point for calculating a warning threshold for CPU load is number of threads per data node * 1.5.

In a scenario where each data node has 2 sockets with 6 cores and hyperthreading, the calculation would be:

2 x 6 x 2 = 24 threads -> 24 * 1.5 = 36