| 28 Sep 2022 |
 Daniel Mawhirter | Things I've done relevant to talosconfig:
$ talosctl gen config "muko-kube" "https://172.22.0.238:6443"
$ talosctl config merge ./talosconfig
$ talosctl config talosctl config node 172.22.0.22 | 00:41:03 |
 Andrew Rynhard | Any console logs you can share? | 00:58:10 |
 Steve Francis | Doesn't he need to pass in the context? | 00:59:31 |
| Steve Francis | muko-kube? | 00:59:39 |
| Daniel Mawhirter | I think the context is ok:
$ talosctl config contexts
CURRENT NAME ENDPOINTS NODES
* muko-kube. 172.22.0.238 172.22.0.22
I don't have console logs from the nodes (I'm looking at them over ipmi), but 172.22.0.22 loops on saying the equivalent of "if this is the first node call talosctl bootstrap"
I've gone ahead and re-generated my configs in case I did something clearly wrong. Re-installing, I'll report back tomorrow with more details if that doesn't clear the issue. Thanks for your time! | 01:12:41 |
| Steve Francis | Let us know! | 01:16:26 |
|  Jason Sievert joined the room. | 01:30:51 |
| Jason Sievert changed their display name from _slack_taloscommunity_U044VT96ZGQ to Jason Sievert. | 01:30:51 |
| Jason Sievert set a profile picture. | 01:30:52 |
 Kris Reeves | if you never actually ran talosctl -n 172.22.0.22 bootstrap , you need to do that (when bootstrapping a control plane from fresh using config) | 02:47:47 |
| Kris Reeves | so the message there is typical | 02:47:58 |
| Kris Reeves | but “authentication handshake failed” is new to me, and i’ve messed up my keys quite a few times recently 😅 | 02:49:02 |
| Daniel Mawhirter | I get the same error when running version or bootstrap | 02:49:14 |
| Kris Reeves | yeah, any command you run that requires the remote server to do something on your behalf will require the grpc connection to first succeed | 02:49:41 |
| Daniel Mawhirter | Ok, keys are my first suspect when I look at this tomorrow then. When you say you’ve messed them up recently, what we’re your symptoms and fix? | 02:51:28 |
| Kris Reeves | mine was a CA error, and the fix was reimage everything because i had overwritten the working thing with one i expected to work but didn’t 😅 | 03:09:16 |
| Kris Reeves | and a few other authorization errors along the way as i was testing things out or working out my workflow | 03:09:32 |
| Kris Reeves | i can’t say if i saw exactly the error you are seeing | 03:09:42 |
| Andrew Rynhard | Have you done apply-config yet? | 03:52:05 |
| Andrew Rynhard | Or how is the node getting the config file? | 03:52:14 |
| Daniel Mawhirter | I’m specifying talos.config as an http url pointing to controlplane.yaml on the kernel command line | 04:22:35 |
 frezbo | i would assume it failed to get the config | 04:49:42 |
| frezbo | would you be able to provide the console logs? | 04:49:53 |
 shax11 | Hi, when doing a interactive install, is there any way to select the disk by id? | 05:30:17 |
| frezbo | no, it's not supported | 05:54:16 |
 Manuel Schmitzberger | I can reproduce the issue on a VM:
rbd -m 10.2.0.10:6789 -p ceph-k8s --id kube --key key ls
did not load config file, using default settings.
2022-09-28T07:30:04.048+0000 7f67b0af2380 -1 Errors while parsing config file!
2022-09-28T07:30:04.048+0000 7f67b0af2380 -1 parse_file: filesystem error: cannot get file size: No such file or directory [ceph.conf]
2022-09-28T07:30:04.048+0000 7f67b0af2380 -1 Errors while parsing config file!
2022-09-28T07:30:04.048+0000 7f67b0af2380 -1 parse_file: filesystem error: cannot get file size: No such file or directory [ceph.conf]
I've to create an empty ceph.conf file (/etc/ceph/ceph.conf). ¯\(ツ)/¯ | 07:31:49 |
 andrey | just to make sure, try connecting to the node in insecure mode to rule out missing machine config: talosctl -n IP disks --insecure
if this fails with an error like 'EOF' or 'connection closed", the node got its configuration
the error is really strange though, but it might happen if the talosconfig certs/CA don't match the Talos one (i.e. you need to use matching talosconfig and controlplane.yaml/worker.yaml) | 09:14:28 |
| Manuel Schmitzberger | & this error is coming from kubelet.
Seems I've to create an empty ceph.conf file on the host (/etc/ceph/ceph.conf).
Is this possible? | 09:21:40 |
|  Shahaf Benita changed their profile picture. | 13:47:55 |
| shax11 | Does this mean the interactive install should not be used with multiple disks because of sd* naming inconsistencies or I'm missing something? | 14:06:34 |
