Back to bug 1025467
| Who | When | What | Removed | Added |
|---|---|---|---|---|
| Chris Pelland | 2013-10-31 18:55:35 UTC | Assignee | nobody | sgotliv |
| QA Contact | acanan | |||
| Sergey Gotliv | 2013-11-03 14:40:26 UTC | Status | NEW | POST |
| Marina Kalinin | 2013-11-04 16:14:05 UTC | Flags | needinfo?(abaron) | |
| Pablo Iranzo Gómez | 2013-11-05 10:39:39 UTC | CC | pablo.iranzo | |
| Lee Yarwood | 2013-11-05 11:20:12 UTC | CC | lyarwood | |
| Sean Cohen | 2013-11-13 09:30:17 UTC | Blocks | 797961 | |
| Sean Cohen | 2013-11-18 13:46:04 UTC | Link ID | RHEV gerrit 10301 | |
| Status | POST | MODIFIED | ||
| Blocks | 797961 | |||
| errata-xmlrpc | 2013-11-26 10:34:17 UTC | Status | MODIFIED | ON_QA |
| Aharon Canan | 2013-11-28 13:56:43 UTC | Status | ON_QA | VERIFIED |
| Charlie | 2013-11-29 02:21:56 UTC | CC | cboyle | |
| Doc Text | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set). Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. |
|||
| Charlie | 2013-11-29 02:24:35 UTC | Doc Text | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set). Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set).Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. |
| Charlie | 2013-11-29 04:29:40 UTC | Doc Text | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set).Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set).Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. |
| Charlie | 2013-11-29 04:33:11 UTC | Doc Text | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set).Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set).Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. |
| Ayal Baron | 2013-12-03 16:01:30 UTC | Flags | needinfo?(abaron) | |
| Charlie | 2013-12-13 00:39:13 UTC | Doc Text | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set).Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns nonzero code, but vgs returns zero even if devices are filtered . The patch introduces public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. | After adding a new LUN to the existing storage domain (both FC and iSCIS BlockSD are affected) hosts (except SPM) became non-operational. VMs started to migrate but mostly failed. The domainMonitor thread might have been using stale lvm filters while running the vgs command after extending of the VG (added PV is missed in the filter set).Since LvmCache doesn't invalidate filters, the volume group is marked as partial and domainMonitor selftest fails (host goes to non-operational status). By design, the filter should be invalidated if cmd returns a nonzero code, but vgs returns zero even if the devices are filtered . The patch introduces a public method for filter invalidation and calls it from the getDevicesVisibility, because getDevicesVisibility is called on all DC hosts after adding a new device but before extendSD. |
| errata-xmlrpc | 2013-12-18 08:01:24 UTC | Status | VERIFIED | RELEASE_PENDING |
| errata-xmlrpc | 2013-12-18 13:58:15 UTC | Status | RELEASE_PENDING | CLOSED |
| Resolution | --- | ERRATA | ||
| Last Closed | 2013-12-18 08:58:15 UTC | |||
| Simon Reber | 2014-03-19 09:44:37 UTC | CC | sreber | |
| Allon Mureinik | 2016-02-10 17:47:09 UTC | oVirt Team | --- | Storage |
Back to bug 1025467