[pve-devel] [PATCH ha-manager 0/6] catch problems when updating cluster FS

[Thomas Lamprecht]

A series which, besides a few small improvements detected during looking at
this, addresses issues when updating pve-cluster.
The highest chance to trigger this is on the current master node with active
services.
The, in short is, that we had a few places where we did not caught errors
resulting from an (temporarily) unavailable cluster FS and thus instantly died,
circumventing our lock logic which should handle this case.

Thomas Lamprecht (6):
  swap native syslog command with HA environment one
  node status: abort update if not quorate
  Env/PVE2: get_node_info: ensure quorate and actual info is used
  lrm: handle an error during service_status update
  CRM: refactor check if state transition to active is ok
  read_resources_config: ensure we got uptodate data

 src/PVE/HA/CRM.pm                  | 40 +++++++++++++++++++++++++-------------
 src/PVE/HA/Config.pm               |  1 +
 src/PVE/HA/Env/PVE2.pm             | 22 +++++++++++----------
 src/PVE/HA/LRM.pm                  | 22 +++++++++++++++++++--
 src/PVE/HA/NodeStatus.pm           |  4 +++-
 src/PVE/HA/Tools.pm                | 15 --------------
 src/test/test-reboot1/log.expect   |  1 +
 src/test/test-shutdown1/log.expect |  1 +
 src/test/test-shutdown2/log.expect |  1 +
 src/test/test-shutdown3/log.expect |  1 +
 src/test/test-shutdown4/log.expect |  1 +
 11 files changed, 67 insertions(+), 42 deletions(-)

-- 
2.11.0