LXC Copy of container with profile different on host and remote fail with Socket Error

LXD
1

Hi,
I try to sync container between 2 hosts.
Profile are differents between source and destination, so I use the -p flag but it fail:

# /snap/bin/lxc copy local:redmine remote:bck-redmine --refresh --mode=push -p default
Error: Failed instance migration: Failed reading migration index header: websocket: close 1006 (abnormal closure): unexpected EOF

lxd monitor display this error on remote:


location: none
metadata:
  context:
    class: websocket
    description: Creating instance
    err: 'Error transferring instance data: Profile not found'
    operation: 7c619393-ee9d-4e65-96e5-4f03947c118e
    project: default
  level: debug
  message: Failure for operation
timestamp: "2022-12-07T11:54:32.61458309+01:00"
type: logging


location: none
metadata:
  class: websocket
  created_at: "2022-12-07T11:54:32.378287067+01:00"
  description: Creating instance
  err: 'Error transferring instance data: Profile not found'
  id: 7c619393-ee9d-4e65-96e5-4f03947c118e
  location: none
  may_cancel: false
  metadata:
    control: d5876ca2a261aa332f6d21640828d62cdde01b822de519473d08d09a6036b0ca
    fs: 64c87cc151aba4249a2f6cb366fd8e0b2fab5b7df307e5c44a7b65fa64fc9ad1
  resources:
    containers:
    - /1.0/containers/bck-redmine
    instances:
    - /1.0/instances/bck-redmine
  status: Failure
  status_code: 400
  updated_at: "2022-12-07T11:54:32.378287067+01:00"
project: default
timestamp: "2022-12-07T11:54:32.614696903+01:00"
type: operation

Strange thing, it fail on 1 host only, I had a second server with the same specific profile (lanprofile) and it success :thinking:

Note:

  • I had to upgrade local LXD from 4.x to 5.8 just before because lxc copy hang (cf my previous question)
  • my backup script was not bullet proof so multiple execution occured in parralell (now corrected)
  • bck-redmine is not visible on remote (try to check in lxd database but don’t find how)

Thanks for your time

3

So both hosts run lxd 5.8?

Thanks

4

Yes:

Client version: 5.8
Server version: 5.8

On both server

5

I just made another test, if I create another container on the same source server with default profile (lxc init test) instead of my custom profile (lanprofile) I can successfully copy it to my remote (/snap/bin/lxc copy local:test remote:bck-test --refresh --mode=push -p default)

The problem seem to be that my -p default flag is not recognised ?

6

I created on destination a copy of default profile named like the used source profile and it works.
The problem is clearly in -p flag

7

Thanks, I’ve assigned this to myself and its on my list to look at.

Thanks for the detail.

1 Like
8

So I tried to reproduce this (unsuccessfully) using LXD 5.9:

Host A:

lxc profile create foo
lxc profile show default | lxc profile edit foo
lxc profile list
+---------+---------------------+---------+
|  NAME   |     DESCRIPTION     | USED BY |
+---------+---------------------+---------+
| default | Default LXD profile | 2       |
+---------+---------------------+---------+
| foo     | Default LXD profile | 1       |
+---------+---------------------+---------+

lxc init images:alpine/3.17 c1 -p foo
lxc config show c1
architecture: x86_64
config:
  image.architecture: amd64
  image.description: Alpine 3.17 amd64 (20221213_13:01)
  image.os: Alpine
  image.release: "3.17"
  image.requirements.secureboot: "false"
  image.serial: "20221213_13:01"
  image.type: squashfs
  image.variant: default
  volatile.apply_template: create
  volatile.base_image: fb6a90c9a45b42c74dbbf2c27551a02a63ecd920016609aaa2758079479b0c0d
  volatile.cloud-init.instance-id: bb76df35-0675-4563-b4a9-3764cd2667b1
  volatile.eth0.hwaddr: 00:16:3e:94:a3:73
  volatile.eth0.name: eth0
  volatile.idmap.base: "0"
  volatile.idmap.next: '[{"Isuid":true,"Isgid":false,"Hostid":1000000,"Nsid":0,"Maprange":1000000000},{"Isuid":false,"Isgid":true,"Hostid":1000000,"Nsid":0,"Maprange":1000000000}]'
  volatile.last_state.idmap: '[]'
  volatile.uuid: 0c2d28db-519d-436d-8c79-4c6530a06774
devices: {}
ephemeral: false
profiles:
- foo
stateful: false
description: ""

Host B:

lxc profile list
+---------+---------------------+---------+
|  NAME   |     DESCRIPTION     | USED BY |
+---------+---------------------+---------+
| default | Default LXD profile | 4       |
+---------+---------------------+---------+

Host A:

lxc copy c1 hostb: --mode=push --refresh
Error: Requested profile "foo" doesn't exist

lxc copy c1 hostb: --mode=push --refresh -p default
lxc copy c1 hostb: --mode=push --refresh -p default

Host B:

lxc config show c1
architecture: x86_64
config:
  image.architecture: amd64
  image.description: Alpine 3.17 amd64 (20221213_13:01)
  image.os: Alpine
  image.release: "3.17"
  image.requirements.secureboot: "false"
  image.serial: "20221213_13:01"
  image.type: squashfs
  image.variant: default
  volatile.apply_template: create
  volatile.base_image: fb6a90c9a45b42c74dbbf2c27551a02a63ecd920016609aaa2758079479b0c0d
  volatile.cloud-init.instance-id: bb76df35-0675-4563-b4a9-3764cd2667b1
  volatile.eth0.hwaddr: 00:16:3e:94:a3:73
  volatile.eth0.name: eth0
  volatile.idmap.base: "0"
  volatile.idmap.next: '[{"Isuid":true,"Isgid":false,"Hostid":1000000,"Nsid":0,"Maprange":1000000000},{"Isuid":false,"Isgid":true,"Hostid":1000000,"Nsid":0,"Maprange":1000000000}]'
  volatile.last_state.idmap: '[]'
  volatile.uuid: 0c2d28db-519d-436d-8c79-4c6530a06774
devices: {}
ephemeral: false
profiles:
- default
stateful: false
description: ""

All seems to be working as expected.

Is your lxc client command also 5.8?
Does using LXD 5.9 fix it?

9

Are you also using different projects?

11

Thanks for your time, I’me not surprised because as I said, on another server with the same profile it’s works fine :confused:
For now, the copy of my default profile on destination with the same name of source profile make the job.
Still don’t updated my servers to 5.9 so I can’t tell you.

I don’t understand your last question (projects ?)

12

I wondered if your instances were in different LXD projects?

13