https://rancher.com/ logo
Title
s

sticky-summer-13450

10/29/2022, 9:41 AM
I find lots of aspects of Harvester's processes completely opaque. For instance, this VM backup has been running for an hour and a quarter, but I have no idea what is happening, why it's so slow, where it is in the progress. There are not even any logs or events I can find in the UI that I might use to glean fragments of information from.
There are other things, like I have a VM which will not start running - it flip-flops between starting and paused. There is nothing in the "logs" for that host. I have no way to investigate why it's not starting.
And the upgrade which I spoke of above. There are no logs available in the UI. Nothing for me to guess why it's stopped after successfully upgrading one node.
Now I love Harvester and what it's doing - it's really amazing, but also terribly frustrating. I really want it to succeed, but it needs to evolve to be less opaque and more open in informing its users what's happening behind the scenes.
Another example where I can see the progress, or lack there of, but I cannot see logs or get any clue why there is no progress.
b

bright-fireman-42144

10/29/2022, 4:06 PM
Yes, I agree. Two things I would like to make this spaghetti mess more reasonable for the 'average' yet somewhat clueful enterprise administrator is better log management in the UI of all the different components with associated resolution/where to look next. Bread crumbs on which rabbit hole to fall down as there is so many moving components under the covers. The other area and this is similar to challenges a lot of people face is an easier to use certificate provisioning process with letsencrypt. I'm just using it as a single node lab to replace an ESXi host I was using and even that doc I sent was unable to uncover what my issue was and I ended up just installing from scratch. I was between phase 0 and phase 1 or their troubleshooting and said screw it... faster for me to rebuild.
s

sticky-summer-13450

10/29/2022, 4:20 PM
I find adding a LetsEncrypt cert is easier than with ESXi. When I tried with ESXi there was lots of
scp
of the fullchain.pem, cert.pem & privkey.pem followed by an
ssh -i -C "services.sh restart"
which completely crashed node. For Harvester I use this script:
#!/bin/bash

set -e

publicCertificate=$(awk 'NF {sub(/\r/, ""); printf "%s\\n",$0;}' /path/to/fullchain.pem)
privateKey=$(awk 'NF {sub(/\r/, ""); printf "%s\\n",$0;}' /path/to/privkey.pem)

cat << EOF | kubectl apply -f - --context <harvester-cluster-context>
apiVersion: <http://harvesterhci.io/v1beta1|harvesterhci.io/v1beta1>
kind: Setting
metadata:
  name: ssl-certificates
value: >-
  {
    "publicCertificate":"${publicCertificate}",
    "privateKey":"${privateKey}"
  }
EOF
b

bright-fireman-42144

10/29/2022, 4:21 PM
oooo... that looks handy! yoink thank you!
­čśâ 1