| | 1 | == !HealthCheck == |
| | 2 | |
| | 3 | The healthcheck.py script must be run in a '''root shell''' as it requires admin privileges for several operations. |
| | 4 | To use this tool, run the following commands from the control node: |
| | 5 | {{{ |
| | 6 | cd /home/gram/gram/src/gram/am/gram |
| | 7 | sudo bash |
| | 8 | source /etc/novarc |
| | 9 | export PYTHONPATH=$PYTHONPATH:/home/gram/gram/src |
| | 10 | python healthcheck.py |
| | 11 | }}} |
| | 12 | |
| | 13 | Healthcheck performs various checks including: |
| | 14 | * Checking the state of the GRAM services |
| | 15 | * Checking the state of the !OpenStack services |
| | 16 | * Checking if the management namespace exists and is properly configured |
| | 17 | * Checking that the compute hosts are reachable |
| | 18 | * Checking that the AM in functioning by doing Allocate-Provision-Delete |
| | 19 | |
| | 20 | If it detects errors, it prints out warnings and attempts to fix them. |
| | 21 | |
| | 22 | Occasionally, after restart, there is an issue with !OpenStack where the management namespace is not instantiated. In this case, healthcheck will attempt to recover the namespace by restarting the Quantum-L3 agent up to 10 times. If this does not work, as a last resort, the admin can tear down the recreate the management and public networks. Healthcheck will prompt to recreate the network. |
| | 23 | |
| | 24 | * Before recreating the networks, the old ones must be removed. |
| | 25 | |
| | 26 | The simplest method is to use the Horizon browser interface to delete the routers and networks. |
| | 27 | Then use healthcheck to recreate these components. |
| | 28 | |
