This message was deleted.

My theory is that for some reason the traffic that is outbound from harvester is sent through some internal ingress- however - when looking at the log from the ingress controller - no error messages are found.

Did you set any

http_proxy

in harvester ? and did you set the

ssl-certificates, addtional CA

?

No this is a clean installation without any configurations. No proxy and no CA certs.

try create an image via URL: https://download.cirros-cloud.net/0.5.2/cirros-0.5.2-x86_64-disk.img

I exec:ed to the harvester pod in the harvester-system namespace and found that it has the same problem when trying to curl an external https url.

If your network is using

http-proxy

, then you need to configure them in Harvester.

I understand but I’m not using a proxy - the nodes have no problems reaching internet but harvester is not able to communicate

How long is the cluster running after installation ?

This is a problem that occurs directly after installing the cluster and I have tried reinstalling the cluster both with 1.2.0 and 1.1.2

please

sudo -i cat /oem/harvester.config

after ssh into Harvester NODE

This is the contents of the config

sudo -i cat /oem/harvester.config
schemeversion: 1
serverurl: ""
token: ion-travesty-shaded
os:
  sshauthorizedkeys:
    ... redacted ...
  writefiles: []
  hostname: harvester-6wzrr
  modules:
  - kvm
  - vhost_net
  sysctls: {}
  ntpservers:
  - <http://sth1.ntp.se|sth1.ntp.se>
  - <http://sth2.ntp.se|sth2.ntp.se>
  - <http://sth3.ntp.se|sth3.ntp.se>
  - <http://sth4.ntp.se|sth4.ntp.se>
  dnsnameservers:
  - 1.1.1.1
  - 1.0.0.1
  wifi: []
  password: agenda-mascot-plane
  environment: {}
  labels: {}
install:
  automatic: true
  mode: create
  managementinterface:
    interfaces:
    - name: eno1
      hwaddr: d0:43:1e:9b:ec:81
    method: dhcp
    ip: ""
    subnetmask: ""
    gateway: ""
    defaultroute: false
    bondoptions: {}
    mtu: 0
    vlanid: 0
  vip: 172.30.10.99
  viphwaddr: ""
  vipmode: static
  forceefi: false
  device: /dev/sda
  configurl: <http://redacted>
  silent: false
  isourl: <http://redacted>
  poweroff: false
  noformat: false
  debug: false
  tty: tty1
  forcegpt: true
  forcembr: false
  datadisk: ""
  webhooks: []
runtimeversion: v1.24.11+rke2r1
rancherversion: v2.6.11
harvesterchartversion: 1.1.2
monitoringchartversion: 100.1.0+up19.0.3
systemsettings: {}
loggingchartversion: 100.1.3+up3.17.7

As a workaround, you may try

download

the image to your PC, then create image via

upload

from the downloaded file.

Yes, it works if I download it to my internal network but not from internet.

cool

My dhcp config had a local search domain which I changed from the default .local to an actual domain name which made everything strange. That was the culprit and the way I discovered what was going on was that I tried to narrow down the problem from within the harvester pod where I tried to curl 1.1.1.1 and it worked but when trying another web address I always got the same IP:s in response. I found that these IP:s was connected to another ISP where we have some other servers so I understood it had something to do with the DNS resolve- not routing or network (strangely though- nslookup and dig all resolved ok). Then I tried to narrow down where we have made configurations that are specific to us (since the harvester installation was a clean one). And eventually found the dhcp config where this name prevented the pods from accessing the internet.