Revision 2c9fb8e2
| b/man/hroller.rst | ||
|---|---|---|
| 37 | 37 |
which set of nodes can be rebooted at the same time while avoiding |
| 38 | 38 |
having both primary and secondary nodes being rebooted at the same time. |
| 39 | 39 |
|
| 40 |
For backends that support identifying the master node (currenlty |
|
| 41 |
RAPI and LUXI), the master node is scheduled as the last node |
|
| 42 |
in the last reboot group. |
|
| 43 |
|
|
| 40 | 44 |
ALGORITHM FOR CALCULATING OFFLINE REBOOT GROUPS |
| 41 | 45 |
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 42 | 46 |
|
| ... | ... | |
| 56 | 60 |
BUGS |
| 57 | 61 |
---- |
| 58 | 62 |
|
| 59 |
The master node should be always the last node of the last group, or anyway |
|
| 60 |
somehow easily identifiable. Right now this is not done. |
|
| 61 |
|
|
| 62 | 63 |
Offline nodes should be ignored. |
| 63 | 64 |
|
| 64 | 65 |
If instances are online the tool should refuse to do offline rolling |
| ... | ... | |
| 73 | 74 |
EXAMPLE |
| 74 | 75 |
------- |
| 75 | 76 |
|
| 76 |
Note that these examples may not for the latest version. |
|
| 77 |
|
|
| 78 | 77 |
Offline Rolling node reboot output |
| 79 | 78 |
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 80 | 79 |
|
| 81 | 80 |
With the default options, the program shows one reboot group per line as |
| 82 | 81 |
a comma-separated list. |
| 82 |
:: |
|
| 83 | 83 |
|
| 84 |
$ hroller |
|
| 84 |
$ hroller -L
|
|
| 85 | 85 |
'Node Reboot Groups' |
| 86 | 86 |
node1.example.com,node3.example.com,node5.example.com |
| 87 | 87 |
node8.example.com,node6.example.com,node2.example.com |
Also available in: Unified diff