Difference between revisions of "Procedure Downtime Declaration"
Andrea.manzi (Talk | contribs) m (moved Procedure Infrastructure Downtime to Procedure Downtime) |
Andrea.manzi (Talk | contribs) |
||
Line 1: | Line 1: | ||
− | Any scheduled | + | Any scheduled downtime must be announced in advance to the user communities. The announcement is done by the [[Role Infrastructure Manager|Infrastructure Manager]], [[Role VO Manager|VO Manager]] or [[Role VRE Manager|VRE Manager]] exploiting the facilities offered on each VO/VRE by the Resource Management portlet. |
+ | |||
+ | [[EcosystemDowntime.png|centre]] | ||
The downtime of the infrastructure can be caused by (1) an infrastructure upgrade or (2) the unavailability of nodes (check below). In this second case, different procedures apply whether the unavailability affects gCube nodes, gLite nodes or Hadoop nodes. In any case, the [[Role Infrastructure Manager|Infrastructure Manager]] must analyse the impact of the nodes unavailability and declare the appropriate infrastructure downtime. | The downtime of the infrastructure can be caused by (1) an infrastructure upgrade or (2) the unavailability of nodes (check below). In this second case, different procedures apply whether the unavailability affects gCube nodes, gLite nodes or Hadoop nodes. In any case, the [[Role Infrastructure Manager|Infrastructure Manager]] must analyse the impact of the nodes unavailability and declare the appropriate infrastructure downtime. |
Revision as of 16:53, 7 December 2011
Any scheduled downtime must be announced in advance to the user communities. The announcement is done by the Infrastructure Manager, VO Manager or VRE Manager exploiting the facilities offered on each VOVirtual Organization;/VREVirtual Research Environment. by the Resource Management portlet.
The downtime of the infrastructure can be caused by (1) an infrastructure upgrade or (2) the unavailability of nodes (check below). In this second case, different procedures apply whether the unavailability affects gCube nodes, gLite nodes or Hadoop nodes. In any case, the Infrastructure Manager must analyse the impact of the nodes unavailability and declare the appropriate infrastructure downtime.
Information about all declared scheduled downtimes is gathered here.
gCube Nodes
The unavailability of gCube nodes must be declared by the Site Manager by sending an email to the SA1 mailing list. The email must contain information about the unavailability duration as well as a list of the affected nodes.
gLite Nodes
The unavailability of gLite nodes must be declared by the Site Manager following the EGI downtime procedure. Site Managers must declare the unavailability using the GOCDB website in the section related to their site.
Hadoop Nodes
The unavailability of Hadoop nodes must be declared by the Site Manager by sending an email to the SA1 mailing list. The email must contain information about the unavailability duration as well as a list of the affected nodes.