[pve-devel] [PATCH docs 1/2] pvecm: Add required -r to rm

[pve-devel] [PATCH docs 1/2] pvecm: Add required -r to rm

[Dominic Jäger]

/etc/corosync/* includes the directory uidgid.d.
Consequentlly, a correct rm call requires -r.

Signed-off-by: Dominic Jäger <d.jaeger@proxmox.com>
---
 pvecm.adoc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pvecm.adoc b/pvecm.adoc
index 4bf2f59..9207571 100644
--- a/pvecm.adoc
+++ b/pvecm.adoc
@@ -414,7 +414,7 @@ Delete the corosync configuration files:
 [source,bash]
 ----
 rm /etc/pve/corosync.conf
-rm /etc/corosync/*
+rm -r /etc/corosync/*
 ----
 
 You can now start the filesystem again as normal service:
-- 
2.20.1

[pve-devel] [PATCH docs 2/2] pvecm: Add output for delnode

[Dominic Jäger]

The output of "pvecm delnode someNode" is "Killing node X". Even though this
only says something about an attempt and not about success, it is not "no
output is returned".

Signed-off-by: Dominic Jäger <d.jaeger@proxmox.com>
---
 pvecm.adoc | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/pvecm.adoc b/pvecm.adoc
index 9207571..220c09f 100644
--- a/pvecm.adoc
+++ b/pvecm.adoc
@@ -327,11 +327,11 @@ After powering off the node hp4, we can safely remove it from the cluster.
 
 ----
  hp1# pvecm delnode hp4
+ Killing node 4
 ----
 
-If the operation succeeds no output is returned, just check the node
-list again with `pvecm nodes` or `pvecm status`. You should see
-something like:
+Use `pvecm nodes` or `pvecm status` to check the node list again. It should
+look something like:
 
 ----
 hp1# pvecm status
-- 
2.20.1

[pve-devel] applied: [PATCH docs 2/2] pvecm: Add output for delnode

[Thomas Lamprecht]

On 22.07.20 12:20, Dominic Jäger wrote:
> The output of "pvecm delnode someNode" is "Killing node X". Even though this
> only says something about an attempt and not about success, it is not "no
> output is returned".
> 
> Signed-off-by: Dominic Jäger <d.jaeger@proxmox.com>
> ---
>  pvecm.adoc | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
>

applied, thanks!

Re: [pve-devel] [PATCH docs 1/2] pvecm: Add required -r to rm

[Thomas Lamprecht]

On 22.07.20 12:20, Dominic Jäger wrote:
> /etc/corosync/* includes the directory uidgid.d.
> Consequentlly, a correct rm call requires -r.

Does leaving this directory left-over causes any issues?

IIRC, I omit the "-r" explicitly as there can be also a directory from
a qdevice or other ones, which we do not want to remove - to level
corosync/* files where all desired to be removed, thus this rm varian.

But, maybe we should rather change it to:
# rm -f /etc/corosync/corosync.conf /etc/corosync/authkey

to make that more explicit

> 
> Signed-off-by: Dominic Jäger <d.jaeger@proxmox.com>
> ---
>  pvecm.adoc | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/pvecm.adoc b/pvecm.adoc
> index 4bf2f59..9207571 100644
> --- a/pvecm.adoc
> +++ b/pvecm.adoc
> @@ -414,7 +414,7 @@ Delete the corosync configuration files:
>  [source,bash]
>  ----
>  rm /etc/pve/corosync.conf
> -rm /etc/corosync/*
> +rm -r /etc/corosync/*
>  ----
>  
>  You can now start the filesystem again as normal service:
> 

Re: [pve-devel] [PATCH docs 1/2] pvecm: Add required -r to rm

[Dominic Jäger]

On Wed, Jul 22, 2020 at 12:31:53PM +0200, Thomas Lamprecht wrote:
> On 22.07.20 12:20, Dominic Jäger wrote:
> > /etc/corosync/* includes the directory uidgid.d.
> > Consequentlly, a correct rm call requires -r.
> 
> Does leaving this directory left-over causes any issues?
When I tested this it was empty. It should contain some optional permissions
for corosync [0].

We later say "(...) delete all remaining files left from the old cluster. This
ensures that the node can be added to another cluster again without problems."

Couldn't those permissions then be active in a potential new cluster?

> IIRC, I omit the "-r" explicitly as there can be also a directory from
> a qdevice or other ones, which we do not want to remove - to level
> corosync/* files where all desired to be removed, thus this rm varian.

Looking at the quote from before, why do we want some connection to the qdevice
to remain?

> But, maybe we should rather change it to:
> # rm -f /etc/corosync/corosync.conf /etc/corosync/authkey

I just tried it without -f:

root@pveA:/etc/corosync# ls
authkey  corosync.conf  uidgid.d
root@pveA:/etc/corosync# rm /etc/corosync/corosync.conf /etc/corosync/authkey
root@pveA:/etc/corosync# ls
uidgid.d

And the node was removed from the cluster as expected.

I can test it with permissions and qdevice, too. Would just have look up how
that actually works.

[0] https://www.mankier.com/8/pcs (ctrl+f for uidgid)

[pve-devel] applied: [PATCH docs 1/2] pvecm: Add required -r to rm

[Thomas Lamprecht]

On 22.07.20 12:20, Dominic Jäger wrote:
> /etc/corosync/* includes the directory uidgid.d.
> Consequentlly, a correct rm call requires -r.
> 
> Signed-off-by: Dominic Jäger <d.jaeger@proxmox.com>
> ---
>  pvecm.adoc | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
>

applied, thanks!