Storage Monitoring and Accounting
Contents
Storage Element Monitoring and Accounting Overview
See also related pages such as
- Monitoring Tools for LCG
- GridPP dCache GIP plugin
- Nagios Plugins
- MonAMI DPM plugin
- MonAMI dCache plugin
- RAL Tier1 Metrics for GridPP
In this page we discuss proposed and actual storage monitoring, both from the site's perspective and the central perspective.
Monitoring and Accounting and Metrics are vaguely related, in the loose sense that accounting often monitors something, like space used, and metrics are often set in terms of measurables which have to be monitored.
Name | Published | DPM | dCache | CASTOR |
---|---|---|---|---|
Space Used per VO | BDII | OK | OK | Ongoing |
Space Avail per VO | BDII | OK | OK | Ongoing |
Number of users | N/A | N/A | N/A | N/A |
Number of active users | N/A | N/A | N/A | N/A |
Number of requests processed | N/A | N/A | Progress | N/A |
Number of files per VO | N/A | N/A | N/A | N/A |
Availability | N/A | N/A | N/A | N/A |
Name | DPM | dCache | CASTOR |
---|---|---|---|
Pool up | ? | ? | ? |
Daemons running | ? | ? | ? |
Certificates expiring | ? | ? | ? |
Implementing monitoring and publishing data
BDII
Publishes things in the Glue SE Schema - so only things within the schema can be published (unless it is extended). It publishes space used and space avaliable, for each Storage Area.
MonAMI
MonAMI monitoring plugins have been written for DPM and dCache. These plugins, along with more generic ones, allow monitoring of these storage services.
R-GMA
Probing
Best way to check whether an SE is running is to "ping" it. There are three levels:
- Normal network (ICMP) ping - measures whether the host is reachable.
- Protocol ping, like srmPing - checks if the server is running.
- Writing a file into the SE, reading it back, and deleting it.
Probing can be used to get a measure of the SE's availability.