Jump to content

Maintenance Schedule: Difference between revisions

From HPCwiki
IA migration §10: rewrite as maintenance schedule (6-monthly downtimes); reporting via How to Get Help; TODO for live dates (via update-page on MediaWiki MCP Server)
 
(4 intermediate revisions by 2 users not shown)
Line 1: Line 1:
Anunna is taken down for planned maintenance — firmware and software updates — roughly twice a year. Typically one downtime is scheduled in April or May, in a week free of teaching, and another in October, to disrupt usage as little as possible.
== Anunna maintenance ==
Maintenance windows allow us to safely perform essential updates and upgrades that cannot be carried out while the cluster is in use. By grouping this work into planned downtimes, we keep Anunna secure, reliable and up to date while minimising unexpected disruption.
 
Anunna is taken down for planned maintenance — firmware and software updates — roughly twice a year. Typically one downtime is scheduled in April or May and another in October in a week free of teaching, to disrupt usage as little as possible.


During a downtime the cluster is unavailable. Jobs are not scheduled to run across a planned downtime, so check the announced dates when planning long jobs.
During a downtime the cluster is unavailable. Jobs are not scheduled to run across a planned downtime, so check the announced dates when planning long jobs.
 
== Maintenance schedule ==
{| class="wikitable"
! Scheduled downtime
! Date
! Planned work
! Unavailable services
|-
| Autumn 2026
| 12–16 October 2026
| OS patching<br>Lustre upgrade to version 2.17
| HPC compute<br>Login nodes<br>Web applications<br>Lustre storage
|-
| Spring 2027
|
|
|
|-
| Autumn 2027
|
|
|
|}
== Announcements ==
== Announcements ==


Planned maintenance is announced in advance — watch the Message of the Day on the [[Portal Overview|Apps Portal]] and the usual support channels.
Planned maintenance is announced in advance — watch the Message of the Day on the [[Portal Overview|Apps Portal]], login nodes and we always send out an announcement in advance over e-mail.
 
<!-- TODO: confirm where maintenance dates are announced (Apps Portal MOTD, mailing list, Teams) and add the next scheduled downtime, or a link to a live maintenance schedule. -->
 
== Reporting problems ==
== Reporting problems ==



Latest revision as of 08:44, 1 September 2026

Anunna maintenance

Maintenance windows allow us to safely perform essential updates and upgrades that cannot be carried out while the cluster is in use. By grouping this work into planned downtimes, we keep Anunna secure, reliable and up to date while minimising unexpected disruption.

Anunna is taken down for planned maintenance — firmware and software updates — roughly twice a year. Typically one downtime is scheduled in April or May and another in October in a week free of teaching, to disrupt usage as little as possible.

During a downtime the cluster is unavailable. Jobs are not scheduled to run across a planned downtime, so check the announced dates when planning long jobs.

Maintenance schedule

Scheduled downtime Date Planned work Unavailable services
Autumn 2026 12–16 October 2026 OS patching
Lustre upgrade to version 2.17
HPC compute
Login nodes
Web applications
Lustre storage
Spring 2027
Autumn 2027

Announcements

Planned maintenance is announced in advance — watch the Message of the Day on the Apps Portal, login nodes and we always send out an announcement in advance over e-mail.

Reporting problems

If something is not working — during or outside maintenance — see How to Get Help.

See also