https://rancher.com/ logo
Join Slack
Powered by
This message was deleted.
# rancher-desktop
a
This message was deleted.
f
The warnings are benign, so there is no hint on why the command failed 😞
Can you check both 
networking.log
and 
background.log
if there is any more information?
q
I'll note that kubernetes is not enabled....
f
Everything looks normal, except the exit code 255
q
Why is it telling me it's trying to start Kubernetes?
f
Where does it tell you it is starting kubernetes? Just in the notification icon menu? I would ignore this for now. The cert import happens way earlier
What happens when you run the command manually LIMA_HOME=~/Library/Application\ Support/rancher-desktop/lima "/Applications/Rancher Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura" shell --workdir=. 0 sudo update-ca-certificates
q
Copy code
jsoref@jsoref-mbp ~ % LIMA_HOME=~/Library/Application\ Support/rancher-desktop/lima "/Applications/Rancher <http://Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura|Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura>" shell --workdir=. 0 sudo update-ca-certificates







jsoref@jsoref-mbp ~ % echo $?
255
Copy code
jsoref@jsoref-mbp ~ % limactl shell 0
FATA[0000] instance "0" is stopped, run `limactl start 0` to start the instance
Copy code
jsoref@jsoref-mbp ~ % limactl list
NAME       STATUS     SSH            CPUS    MEMORY    DISK      DIR
0          Stopped    127.0.0.1:0    4       4GiB      100GiB    ~/.lima/0
alpine     Stopped    127.0.0.1:0    4       4GiB      100GiB    ~/.lima/alpine
default    Stopped    127.0.0.1:0    4       8GiB      100GiB    ~/.lima/default
oh
those are the wrong instances 🙂
Copy code
jsoref@jsoref-mbp ~ % LIMA_HOME=~/Library/Application\ Support/rancher-desktop/lima "/Applications/Rancher <http://Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura|Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura>" list
NAME    STATUS     SSH                VMTYPE    ARCH       CPUS    MEMORY    DISK      DIR
0       Running    127.0.0.1:51304    vz        aarch64    2       6GiB      100GiB    ~/Library/Application Support/rancher-desktop/lima/0
Copy code
jsoref@jsoref-mbp ~ % LIMA_HOME=~/Library/Application\ Support/rancher-desktop/lima "/Applications/Rancher <http://Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura|Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura>" shell 0 uname -a
jsoref@jsoref-mbp ~ % echo $?
255
f
You are using VZ emulation, so VDE networking isn't being used anyways
I'm confused why you have a 
~/.lima/0
directory; this looks like you have been messing up/missing the 
LIMA_HOME
setting
Maybe just stop Rancher Desktop and restart and see if that fixes it?
Do you know what you did that broke it?
q
"I wasn't touching it!"
(lemme find that "end user" hat of mine)
why wouldn't I have 
0
directory?
f
If you run 
limactl
commands then you may want to set 
LIMA_HOME
in your shell, so that you talk to the right instance
q
I was actually supposed to do some lima/rancher work today, but I hadn't gotten to it yet 🙂
f
Rancher Desktop is completely segregated from Lima and doesn't share any data directories; nothing under 
~/.lima
is related to Rancher Desktop
q
anyway, these are all of my instances
Copy code
jsoref@jsoref-mbp 0 % limactl list
NAME       STATUS     SSH            VMTYPE    ARCH       CPUS    MEMORY    DISK      DIR
0          Stopped    127.0.0.1:0    qemu      aarch64    4       4GiB      100GiB    ~/.lima/0
alpine     Stopped    127.0.0.1:0    qemu      aarch64    4       4GiB      100GiB    ~/.lima/alpine
default    Stopped    127.0.0.1:0    qemu      aarch64    4       8GiB      100GiB    ~/.lima/default
jsoref@jsoref-mbp 0 % LIMA_HOME=~/Library/Application\ Support/rancher-desktop/lima "/Applications/Rancher <http://Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura|Desktop.app/Contents/Resources/resources/darwin/lima/bin/limactl.ventura>" list
NAME    STATUS     SSH                VMTYPE    ARCH       CPUS    MEMORY    DISK      DIR
0       Running    127.0.0.1:63162    vz        aarch64    2       6GiB      100GiB    ~/Library/Application Support/rancher-desktop/lima/0
After restarting Rancher Desktop.app, everyone is happy:
oh
the reason I have a ~/.lima is because I was actually using lima at one point for various things
some involved rd not working, others involved trying to prove that rd wasn't relevant to a problem
f
That's fine, but it is suspicious that you have an instance called 
0
q
what's your instance called?
f
0
, but Rancher Desktop will not create it under 
~/.lima
q
Copy code
jsoref@jsoref-mbp 0 % ls -hal ~/.lima/0
total 1637624
drwx------  10 jsoref  staff   320B Jun 13  2022 .
drwx------   6 jsoref  staff   192B Mar 22 18:36 ..
-rw-r--r--   1 jsoref  staff   578M Jun 13  2022 basedisk
-rw-r--r--   1 jsoref  staff   188M Jun 13  2022 cidata.iso
-rw-r--r--   1 jsoref  staff    34M Jun 13  2022 diffdisk
srwxr-xr-x   1 jsoref  staff     0B Jun 13  2022 ha.sock
-rw-r--r--   1 jsoref  staff   5.5K Jun 13  2022 ha.stderr.log
-rw-r--r--   1 jsoref  staff   146B Jun 13  2022 ha.stdout.log
-rw-r--r--   1 jsoref  staff    17K Jun 13  2022 lima.yaml
-rw-r--r--   1 jsoref  staff    31K Jun 13  2022 serial.log
f
So you must have created it yourself, maybe by copying a command from the log file, but without setting the correct 
LIMA_HOME
value
q
odds of me trying to copy&paste some commands to do things? very high
f
Anyways, that's why I strongly discourage people from using 
limactl
with Rancher Desktop; use 
rdctl shell
instead, which sets things up correctly
Bonus points: it also works on Windows
q
and yet, you gave me a command w/ limactl 😜
f
Yes, but including 
LIMA_HOME
...
Anyways, no rules without exceptions
q
anyway, it was from a year ago, did rdctl even exist then?
f
If you break things, you get to keep all the pieces, and all that 😛
Glad that it works now, so let's blame it on a fluke in the flux capacitor
q
oh sure
less confusion ahead?
Copy code
jsoref@jsoref-mbp 0 % limactl delete 0
INFO[0000] The qemu driver process seems already stopped
INFO[0000] The host agent process seems already stopped
INFO[0000] Removing *.pid *.sock under "/Users/jsoref/.lima/0"
INFO[0000] Removing "/Users/jsoref/.lima/0/ha.sock"
INFO[0000] Deleted "0" ("/Users/jsoref/.lima/0")
(those are from apr/jul of this year)
f
Also gives you back a small amount of your disk space
You may want to delete the other instances too, if you don't need them
q
i have no idea how much
Copy code
jsoref@jsoref-mbp 0 % du -shc ~/.lima
3.3G	/Users/jsoref/.lima
3.3G	total
meh
ok, now i should try to run a 
docker pull
while trying to use tcpdump to capture the logs of it failing miserably
f
The current networking has some serious performance issues, and (surprise) it may be DNS related...
q
...
f
So I have high hopes for switching to 
user-v2
networking in the future
q
my problems were present in lima as well
f
Is this about the 10-15 second delay before it starts pulling the image?
q
and i think i managed to trigger them in a vm using UTM
nah, it's about GCR.io images never successfully pulling
works fine from other places, but not our office
so it shouldn't be a rancher problem
but google and our isp have pointed fingers at eachother
not an enjoyable experience
f
oh, then i don't know. Had the same thing the other day with ghcr.io, but after a day it worked again fine, so 🤷
q
this dates to december at the latest possibly many moons prior
lemme fight MS first
(you know things are bad when I'd rather deal w/ MS)
f
Anyways, it is probably DNS (it invariably is DNS if you look long enough)
q
my dns peers will not be happy to say this
but, yeah, it's quite possible that google has some bad dns discovery for us and is giving us a dumb ip address resulting in lousy performance
that more or less appears to be the problem
oh, wait
f
If you are not using VPN etc, then you can try to disable the hostResolver in your 
override.yaml
file and specify DNS explicitly
q
does RD know to do ipv4 queries over ipv4 and ipv6 queries over ipv6?
f
It should only do IPv4 queries; it does not really work with IPv6
q
ok
f
It's been a while that I looked into it, and I hope we replace everything with gvisor before I have to look again
q
hah
f
The new gvisor-based "networking tunnel" on Windows seems to have fixed most of our existing networking issues on Windows (except for proxy issues)
🙌 1
q
so... from my 90+mb tcpdump, it looks like all of my traffic to google is ipv6
even dns....
476 Views
Open in Slack
PreviousNext
Powered by