[pve-devel] [PATCH docs v2] pveceph: document cluster shutdown

[Aaron Lauterer]

thanks for the review. one comment inline

On  2024-05-23  14:23, Alexander Zeidler wrote:
> On Wed, 2024-05-22 at 10:33 +0200, Aaron Lauterer wrote:
>> Signed-off-by: Aaron Lauterer <a.lauterer at proxmox.com>
>> ---
>> changes since v1:
>> * incorporated suggested changes in phrasing to fix grammar and
>>    distinguish the steps on how to power down the nodes better
>>
>>   pveceph.adoc | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 50 insertions(+)
>>
>> diff --git a/pveceph.adoc b/pveceph.adoc
>> index 089ac80..04bf462 100644
>> --- a/pveceph.adoc
>> +++ b/pveceph.adoc
>> @@ -1080,6 +1080,56 @@ scrubs footnote:[Ceph scrubbing {cephdocs-url}/rados/configuration/osd-config-re
>>   are executed.
>>   
>>   
>> +[[pveceph_shutdown]]
>> +Shutdown {pve} + Ceph HCI cluster
>> +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> +
>> +To shut down the whole {pve} + Ceph cluster, first stop all Ceph clients. This
> Rather s/This/These/ ?
> 
>> +will mainly be VMs and containers. If you have additional clients that might
>> +access a Ceph FS or an installed RADOS GW, stop these as well.
>> +Highly available guests will switch their state to 'stopped' when powered down
>> +via the {pve} tooling.
>> +
>> +Once all clients, VMs and containers are off or not accessing the Ceph cluster
>> +anymore, verify that the Ceph cluster is in a healthy state. Either via the Web UI
>> +or the CLI:
>> +
>> +----
>> +ceph -s
>> +----
>> +
>> +Then enable the following OSD flags in the Ceph -> OSD panel or the CLI:
> For style consistency: **Ceph -> OSD panel**
> 
> Maybe: s/or the CLI/or via CLI/
> 
>> +
>> +----
>> +ceph osd set noout
>> +ceph osd set norecover
>> +ceph osd set norebalance
>> +ceph osd set nobackfill
>> +ceph osd set nodown
>> +ceph osd set pause
> Maybe sort alphabetically as in the UI.

I don't think this is a good idea. The order is roughly going from 
"should be set" to "would be good if set". While it would not be 
affected in this case, as p comes after n, but still is an example that 
a pure alphabetical order by default can be problematic: pause will be 
set last, as is halts any IO in the cluster.

With that in mind, I realized that the sorting in the unset part should 
be reversed.

> 
>> +----
>> +
>> +This will halt all self-healing actions for Ceph and the 'pause' will stop any client IO.
> Perhaps state the goal/result beforehand, e.g.:
> Then enable the following OSD flags in the **Ceph -> OSD panel** or via CLI,
> which halt all self-healing actions for Ceph and 'pause' any client IO:
> 
>> +
>> +Start powering down your nodes without a monitor (MON). After these nodes are
>> +down, continue shutting down hosts with monitors on them.
> Since the continuation is not meant/true for "hosts with monitors":
> s/continue/continue by/
> 
> Maybe: s/hosts/nodes/
> 
>> +
>> +When powering on the cluster, start the nodes with Monitors (MONs) first. Once
> s/Monitors/monitors/
> 
>> +all nodes are up and running, confirm that all Ceph services are up and running
>> +before you unset the OSD flags:
> Maybe stay with either enable/disable or set/unset.
> 
> s/flags:/flags again:/
> 
>> +
>> +----
>> +ceph osd unset noout
>> +ceph osd unset norecover
>> +ceph osd unset norebalance
>> +ceph osd unset nobackfill
>> +ceph osd unset nodown
>> +ceph osd unset pause
> Above mentioned sorting.
> 
>> +----
>> +
>> +You can now start up the guests. Highly available guests will change their state
>> +to 'started' when they power on.
>> +
>>   Ceph Monitoring and Troubleshooting
>>   -----------------------------------
>>   
> 
> 
> 
> _______________________________________________
> pve-devel mailing list
> pve-devel at lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
> 
>