# Downtimes

Managing and consulting downtimes

## Definition

A downtime is a period of time for which a service is declared to be inoperable. Downtimes may be scheduled (e.g. for software/hardware upgrades), or unscheduled (e.g. power outages). The Configuration Database stores the following information about downtimes (non exhaustive list):

• The downtime classification (Scheduled or unscheduled)
• The severity of the downtime
• The date at which the downtime was added
• The start and end of the downtime period
• A description of the downtime
• The entities affected by the downtime

## Manipulating downtimes

### Viewing downtimes

There are different pages in the Configuration Database where downtimes are listed:

• Active & Imminent, linked from the main menu, that allows users to see currently active downtimes and downtimes planned in the coming weeks.
• Downtime Calendar, linked from the main menu, that allows users to view and filter all downtimes.
• Site details, where all the downtimes associated to the site are listed
• Service endpoint details, where all the downtimes associated to the service endpoint are listed.
• Service group details, where all the downtimes associated to the service group are listed.

Each downtime has its own page providing details, accessible by clicking on the Downtime Id link or similar in downtime listing pages.

### Subscribing to downtimes

The EGI Operations Portal provides a publicly-accessible page allowing to view and filter downtimes: Operations Portal.

Authenticated users can subscribe to downtimes affecting sites selected using a filter. The downtimes notifications can be sent by email, RSS and iCal, allowing to easily integrate with your calendar.

Provided you have proper permissions (check the permissions matrix section), you can add a downtime by clicking on the Add Downtime link in the sidebar.

This is done in 2 steps:

1. Enter downtime information
2. Specify the full list of impacted services in case there is more than one or select an site to select all the sites associated services.

• All dates have to be entered in UTC or using the Site Timezone.
• A downtime can be retrospectively added if its start date is less than 48h in the past (giving a 2 day window to add).
• downtime classification (scheduled/unscheduled) is determined automatically (see Scheduled or unscheduled section)

### Editing downtime information

• To edit a downtime, simply click the edit link on top of the downtime’s details page.
• A downtime can be retrospectively updated if its start date is less than 48h in the past (giving a 2 day window to modify).
• Note there are limitations to downtime editing, especially if it has already started, or is due to start in the next 24hrs or is finished. See downtime shortening and extension section for more details.

### Removing downtimes

To delete a downtime, simply click the delete link on top of the downtime’s details page. For integrity reasons, it is only possible to remove downtimes that have not started.

## Good practices and further understanding

### Scheduled or unscheduled

Depending on the planning of the intervention, downtimes can be:

• Scheduled: planned and agreed in advance
• Unscheduled: planned or unplanned, usually triggered by an unexpected failure or at a short term notice

EGI defines precise rules about what should be declared as scheduled or unscheduled, based on how long in advance the downtime is declared. These rules are described in MAN02 Service intervention management and are enforced as follows:

• All downtimes declared less than 24h in advance will be automatically classified as UNSCHEDULED
• All other downtimes will be classified as SCHEDULED

#### Notes

• A downtime can be retrospectively declared and/or updated if its start date is less than 48h in the past (giving a 2 day window to add/modify).
• Although 24h in advance is enough for the downtime to be classified as SCHEDULED, it is good practice to declare it at least 5 working days before it starts.

### WARNING or OUTAGE?

When declaring a downtime, you will be presented the choice of a “severity”, which can be either WARNING or OUTAGE. Please consider the following definitions:

• WARNING means the resource is considered available, but the quality of service might be degraded. Such downtimes generate notifications, but are not taken into account by monitoring and availability calculation tools. In case of a service failure during the WARNING period an OUTAGE downtime has to be declared, cancelling the rest of the WARNING downtime.

• OUTAGE means the resource is considered as unavailable. Such downtimes will be considered as in maintenance by monitoring and availability calculation tools.

### Downtime shortening and extension

Limitation rules to downtime extensions are enforced as follows:

• Scheduled downtimes due to start in 24 hours cannot be edited in any way, nor deleted.
• Other downtimes that have not yet started can be edit and deleted.
• They can be shortened or moved, i.e. They can be edited such that:
• Both start and end time are still in the future
• The duration remains the same or is decreased
• Ongoing downtimes can not be deleted.
• A downtime cannot be edited once it has finished, nor can a new downtime be added more than 48 hours into the past.

If for any reason a downtime already declared needs to be extended, the procedure is to add another adjacent downtime, before or after.