Hello all! How's everyone!? I'm working through a...
# umh-support
t
Hello all! How's everyone!? I'm working through a setup of UMH on an AWS EC2 instance - Ubuntu 22.04 LTS, but receiving this error message right toward the end... Creating UMH namespace āŒ Error: Failed to create namespace united-manufacturing-hub. šŸ’” Tip: Check the logs for any error messages. Event dispatched: 507028a9-7f87-4c69-90ae-8dbc8c662cfa ā„¹ļø For more details, check the logs at /tmp/mgmt-installer/mgmt_install.log ā“ Do you want to display the log file contents? (Y/n):
j
by any chance, did the script before that try 5 tiems to install k3s unsuccessfully, but then proceeded anyway ot install the UMH namespace?
then likely k3s installation failed
but it still proceeded
can you provide me with more logs above what you wrote?
(this bug will be fixed today, but we still need to ivnestiate what is causing the failed k3s installation in the first place)
just tried it with a fresh ubuntu 24.04 and it worked. but I was usign the hetzner ubuntu image, as we have learned yesterday there are differences between cloud images... it would be really great if you could provide the full log, then I could troubleshoot better
t
I'll rebuild and try again to recreate the errors.
j
thank you ā¤ļø
t
šŸ‘
j
jsut tried with a ubuntu 22.04 image as well, and it works. maybe it could be related to the network issue that I am fixing at the moment: the current timeouts all expect a server with a gigabit connection in europe, and if you have a slow connection or a server somewhere else, it could happen that the timeouts are too aggressive. increasing them at the moment
or because of EC2 ubuntu :/
t
What size machine did you launch on?
j
at least 4 cpus and 16GB ram
(also adjsuting the install script to enforce it)
t
Gosh that's expensive šŸ’ø 🤣
j
its incredible expensive, but only on Azure and AWS lol
at hetzner we pay like 30 eur / month for 8cpu and 16GB ram
t
Nice!
j
but hetzner is only in europe and also now in the US, but not in south east asia
but maybe there is something comparable?
t
Maybe I need to try Alicloud
Or get on the AWS partner program and get me some credits! šŸ™‚
j
check ovhcloud
here are the spot instances:
should cost around 70 SGD
/month
t
I'll have a look
I just rebuilt the server, let's try again.
j
also I am just testing the fixes about bad internet connectivity to our servers, should be released in 2-3h
t
Can I install from the instance on the management console, or start a new one?
j
delete the old instance in the management console, and create a new install command
can sue the same VM instance though
t
Install underway
K3s installed correctly
Hmmm, it looks like it's looping through repeatedly trying to install k3s, then uninstalls Kubernetes and trys again
k3s is installed successfully. šŸ”§ Waiting for node to be ready... āœ… Node is ready. šŸ“¢ Validating k3s installation... šŸ” Checking if addons.k3s.cattle.io CRD is installed... āœ… addons.k3s.cattle.io CRD is installed. šŸ” Checking if kube-system namespace has pods running... No resources found in kube-system namespace. šŸ”§ kube-system namespace does not have pods running. Retrying in 5 seconds... šŸ”§ kube-system namespace does not have pods running. Retrying in 5 seconds... šŸ”§ kube-system namespace does not have pods running. Retrying in 5 seconds... šŸ”§ kube-system namespace does not have pods running. Retrying in 5 seconds... šŸ”§ kube-system namespace does not have pods running. Retrying in 5 seconds...
kube-system namespace does not have pods running. Retrying k3s installation... šŸ“¢ Executing RHEL/generic uninstall script āœ… Kubernetes uninstalled successfully.
Seems like it's stuck on Kubernetes
Failed again
@Jermuk Here you go.
j
alright, I know hwat the problem is
t
Of course you do! Wizard šŸ™‚
šŸ§™ā€ā™‚ļø
j
its the bug I just fixed in our development setup šŸ˜„
t
I mean that as a compliment.
j
after trying 5 times, it uninstalled kubernetes but does not abort the script with "could not install k3s" and instead tries to continue (which fails as kubernetes does not exist)
the fix is to wait longer, and abort if necessary with an erro that k3s was not installed
t
Yeah, I watched that happening.
That's something I do?
j
ah no, its what I need to implement šŸ™‚
t
Ah ok, so I'll leave it for awhile.
j
currently trying it out on ubuntu 22.04 with 4 cores and 16gb ram
t
@Jermuk I'm up and running on Ubuntu 24.4 now.
If I wanted to say, access the data that's being stored in the TimescaleDB from running PackML, could I? Maybe for displaying some analytics functionality?
j
yes, that it correct! data from the PackML simulator is fetched via protocol converter and sent to the UNS. From there on, it is automatically stored in TimescaleDB and can be visualized with Grafana (but also any other SQL tool like PowerBI)
2 Views