Bug #766

The scheduler can pick a non working Host

Added by Carlos Martín almost 10 years ago. Updated about 9 years ago.

Status:ClosedStart date:08/08/2011
Priority:NormalDue date:
Assignee:Ruben S. Montero% Done:

0%

Category:Scheduler
Target version:Release 3.6
Resolution:fixed Pull request:
Affected Versions:OpenNebula 3.0

Description

The scheduler filters the existing hosts to get the ones with a status less than 3, including the ones in INIT or MONITORING state.

A Host in MONITORING state may be chosen to deploy a VM if there is old monitoring data in it, but the host may be unreachable and end in ERROR state.

Associated revisions

Revision f162b2c4
Added by Carlos Martín almost 10 years ago

Bug #766: Scheduler only filters Hosts in MONITORED state

Revision 0caef957
Added by Carlos Martín almost 10 years ago

Revert "Bug #766: Scheduler only filters Hosts in MONITORED state"

This reverts commit f162b2c4a97a14f3057f120948bd6cacaf74e7a9.

Revision dabd5702
Added by Ruben S. Montero about 9 years ago

bug #766: New monitoring states for the Hosts

Revision 025a79a6
Added by Ruben S. Montero about 9 years ago

bug #766: Scheduler now uses successfull monitored hosts

Revision 734d64c9
Added by Ruben S. Montero about 9 years ago

bug #766: Update OCA (Ruby & Java) to make use of the new monitoring state. Update onehost command.

Revision e5179a95
Added by Ruben S. Montero about 9 years ago

bug #766: Some Sunstone/Ozones strings for the new states

Revision 0094bfa9
Added by Ruben S. Montero about 9 years ago

feature #766: Update host.xsd with a comment for the host states

History

#1 Updated by Carlos Martín almost 10 years ago

  • % Done changed from 0 to 100

#2 Updated by Ruben S. Montero almost 10 years ago

  • Status changed from New to Closed
  • Resolution set to fixed

#3 Updated by Carlos Martín almost 10 years ago

  • Status changed from Closed to Assigned
  • % Done changed from 100 to 0
  • Resolution deleted (fixed)

The previous commit didn't solve the issue, because a host in MONITORED state changes periodically to MONITORING in each monitoring cycle. With a small monitoring timer, the scheduler couldn't find any host in MONITORED.

We will have to implement one of the following options:

  • Automatically disable a Host after N unsuccessful monitor events.
  • Introduce a new state, to differentiate a host being monitored to refresh its info (valid candidate for deployment), or to recover from an ERROR state.

#4 Updated by Ruben S. Montero almost 10 years ago

  • Target version deleted (Release 3.0)

#5 Updated by Ruben S. Montero almost 10 years ago

  • Target version set to Release 3.4

#6 Updated by Ruben S. Montero almost 10 years ago

  • Status changed from Assigned to New
  • Assignee deleted (Carlos Martín)
  • Affected Versions OpenNebula 3.0 added

#7 Updated by Ruben S. Montero over 9 years ago

  • Target version deleted (Release 3.4)

#8 Updated by Ruben S. Montero about 9 years ago

  • Target version set to Release 3.6

#9 Updated by Ruben S. Montero about 9 years ago

  • Status changed from New to Closed
  • Assignee set to Ruben S. Montero
  • Resolution set to fixed

Fixed now in master, ready for 3.6

Also available in: Atom PDF