Closed Ne02ptzero closed 4 years ago
Do you have a node named XXX
or is it a non-existing node?
XXX
is just me replacing the actual node name. But yes, the node do exist!
I can't reproduce this. Please could you provide more details about your situation? Debug log of the server which is handling the request would be useful. Or even better, a step-by-step reproducer.
I can't provide a step-by-step reproducer, since this is happening everytime, I don't know that triggers the bug.
Client log
Error: Delete https://10.33.0.16:8443/internal/cluster/accept: http: server gave HTTP response to HTTPS client
Local Server log:
DBUG[01-24|15:53:56] Handling user= method=GET url=/1.0 ip=@
DBUG[01-24|15:53:56] Handling method=DELETE url=/1.0/cluster/members/XXX ip=@ user=
DBUG[01-24|15:53:56] Redirect member delete request to 10.33.0.16:8443
I can't see any log related to the request on 10.33.0.16
I can, however, provide an overview of my cluster:
10.33/16
and a specific VLAN. 3 of those nodes are databases member.10.32/16
and a specific VLAN. Those nodes do not have database on them.All the nodes are running 3.19 (master)
. I actually don't know if this bug was introduced in 3.19
, since it's the first time I had to remove a cluster member.
Please can you try to submit the directly request to 10.33.0.16
and see if it shows more insightful logs? It should suffices to point your lxc
client to it.
I was able to reproduce this. I will fix it in the next days.
On LXD
3.19
:With
--debug