|
| 1 | +Frequently Asked Questions |
| 2 | +========================== |
| 3 | + |
| 4 | +This section collects a number of frequently asked questions |
| 5 | +concerning the administration of a StratusLab cloud infrastructure. |
| 6 | +If there are other questions that you would like to see treated here, |
| 7 | +please contact StratusLab support. |
| 8 | + |
| 9 | +How to restart VMs after a power outage? |
| 10 | +---------------------------------------- |
| 11 | + |
| 12 | +After the power outage the nodes hosting the virtual machines must be |
| 13 | +restarted. If this has not happened automatically, ensure that all of |
| 14 | +the nodes have rebooted. |
| 15 | + |
| 16 | +Connect to each of the nodes to ensure that all of the iSCSI mounts |
| 17 | +are present. |
| 18 | + |
| 19 | +:: |
| 20 | + # iscsiadm -m session |
| 21 | + |
| 22 | +If the links are not correctly mounted (bad links will blink in the |
| 23 | +list), then the easiest way to recover this is simply to reboot the |
| 24 | +machine again. |
| 25 | + |
| 26 | +Once you have verified all of the iSCSI links on all of the nodes, |
| 27 | +then you must log into the frontend machine as `oneadmin`. As this |
| 28 | +user, you can then force OpenNebula to restart all of the machines in |
| 29 | +the `unknown` state. |
| 30 | + |
| 31 | +The commands for this (assuming that you're logged into the frontend |
| 32 | +as `root`), are: |
| 33 | + |
| 34 | +:: |
| 35 | + # su - oneadmin |
| 36 | + $ for vm in `onevm list all |grep unk | awk ‘{print $1}’`; \ |
| 37 | + do; \ |
| 38 | + onevm restart $vm; \ |
| 39 | + done |
| 40 | + |
| 41 | +After this, all of the VMs should return to the `running` state and |
| 42 | +again be accessible from the network. Any machines that don't restart |
| 43 | +are probably lost and will have to be killed and a new instance |
| 44 | +created. |
| 45 | + |
0 commit comments