EGI Workload Manager Availability and Continuity Plan
Main | EGI.eu operations services | Support | Documentation | Tools | Activities | Performance | Technology | Catch-all Services | Resource Allocation | Security |
Documentation menu: | Home • | Manuals • | Procedures • | Training • | Other • | Contact ► | For: | VO managers • | Administrators |
Back to main page: Services Availability Continuity Plans
Introduction
This page reports on the Availability and Continuity Plan for EGI Workload Manager - DIRAC4EGI and it is the result of the risks assessment conducted for this service: a series of risks and treats has been identified and analysed, along with the correspondent countermeasures currently in place. Whenever a countermeasure is not considered satisfactory for either avoiding or reducing the likelihood of the occurrence of a risk, or its impact, it is agreed with the service provider a new treatment for improving the availability and continuity of the service. The process is concluded with an availability and continuity test.
Last | Next | |
---|---|---|
Risks assessment | 2019-01-10 | 2020 Jan |
Av/Co plan and test |
Performances
The performances reports in terms of Availability and Reliability are produced by ARGO on an almost real time basis and they are also periodically collected into the Documentation Database.
In the OLA it was agreed the following performances targets, on a monthly basis:
- Availability: 99%
- Reliability 99%
Over the past years, DIRAC4EGI hadn't particular Av/Co issues highlighted by the performances that need to be further investigated.
Risks assessment and management
For more details, please look at the google spreadsheet. We will report here a summary of the assessment.
Risks analysis
Risk id | Risk description | Affected components | Established measures | Risk level | Expected duration of downtime / time for recovery | Comment |
---|---|---|---|---|---|---|
1 | Service unavailable / loss of data due to hardware failure | All the service components | Database protection with regular backups. Redundant Configuration database. Snapshots of virtual machines hosting DIRAC4EGI services | Medium | In case an instance must be instantiated from backups can take up to two-three working hours.
In case latest backups are not available some data must be re-generate from the central accounting repositories and this may take up to two hours. |
the measures already in place are considered satisfactory and risk level is acceptable |
2 | Service unavailable / loss of data due to software failure | All the service components | Backup of the MySQL and Configuration databases | Low | One to two working hours | the measures already in place are considered satisfactory and risk level is acceptable |
3 | service unavailable / loss of data due to human error | All the service components | Backup of the MySQL and Configuration databases | Low | Two to Three working hours. | the measures already in place are considered satisfactory and risk level is acceptable |
4 | service unavailable for network failure (Network outage with causes external of the site) | All the service components | Geographically distributed redundant Configuration Service. Redundant failover Request Management Service. | Low | Close to zero, less than one hour. | the measures already in place are considered satisfactory and risk level is acceptable |
5 | Unavailability of key technical and support staff (holidays period, sickness, ...) | Resources management. User support. Security infrastructure components | Automation of synchronization with BDII, VOMS, GocDB information indices. Automated resource monitoring service | Low | Within the OLA targets for operational actions
Longer periods in case of bugs or maintenance (one week) because not all the personnel can develop patches to the code. |
the measures already in place are considered satisfactory and risk level is acceptable |
6 | Major disruption in the data centre. Fire, flood or electric failure for example | All the service components | Backup of the MySQL and Configuration databases | Low | 1-2 weeks, the time to deploy recover operational status at CESGA or the service to another resource centre partner of the NGI | the measures already in place are considered satisfactory and risk level is acceptable |
7 | Major security incident. The system is compromised by external attackers and needs to be reinstalled and restored. | All the service components | Backup of the MySQL and Configuration databases | Low | 1-2 work hours. In case new host certificates are required, up to 2 days. | the measures already in place are considered satisfactory and risk level is acceptable |
8 | (D)DOS attack. The service is unavailable because of a coordinated DDOS. | All the service components | Limited service queries queues avoiding dangerous overloading of the service components. Automatic service restart after going down due to an overload. | Low | Depending on the attack, few hours maximum | the measures already in place are considered satisfactory and risk level is acceptable |
Outcome
The level of all the identified risks is acceptable and the countermeasures already adopted are considered satisfactory
Availability and Continuity test
The proposed A/C test will focus on a recovery scenario: the service is supposed to have been disrupted and needs to be reinstalled from scratch. Typically this covers the risks 1,2, and 7. The last backup of the data will be used for restoring the service, verifying how much information will be lost, and the time spent will be measured.
Performing this test will be useful to spot any issue in the recovery procedures of the service.
Test details
references in the RT ticket 14308
10:45 CEST -> poweroff of the original machine (accounting-devel-next) 11:05 CEST -> Power on for the template (minimal) machine for recovery 12:00 CEST -> Data recovery completed 12:10 CEST -> Boot restoration and miscelaneus comprobations. 12:15 CEST -> Recovered machine rebooted and services on line
The recovery took about 70 minutes.
Test outcome
The test can be considered successful: the service can be restored in few time and there are no loss of data. Even if the service is not available for few hours or a day, this can be considered acceptable: the portal is used for displaying and collecting accounting information, the other infrastructure services are not depending on it. Only infrastructure/operations centres/resource centres/VOs managers would suffer of a temporary disruption of the service.
Revision History
Version | Authors | Date | Comments |
---|---|---|---|
Alessandro Paolini | 2018-05-03 | first draft, discussing with the provider | |
Ivan Diaz, Alessandro Paolini | 2018-05-28 | added a paragraph about HA configuration; added information about the test | |
Alessandro Paolini | 2018-08-15 | added the paragraph Test Outcome, plan finalised | |