Closed datho7561 closed 1 year ago
crc setup worked fine for me but crc start
hangs in the same stage. F38 as well.
It looks like this is the issue: https://github.com/systemd/systemd/issues/27953
Workaround: systemctl restart libvirtd.service
Seems like libvirtd was disabled for me. After enabling it, crc
seems to be working
A restart might not work, like a fresh install. In that case libvirtd
might not have been marked as enable and won't (re)start; Job for libvirtd.service canceled.
.
Better to do the following in that case:
$ sudo systemctl enable --now libvirtd.service
I've been fighting this for the past few hours when I tried updating crc to the latest release. I completely uninstalled libvert daemon and qemu-kvm and reinstalled, but that fixed nothing. Finally got the it working after starting and enabling virtnetworkd, although I can't quite say for sure this is what fixed it. I am also pretty sure that libvirtd doesn't actually have anything to do with running the VM, because I can clearly see it isn't running from the Cockpit console?
The following was in the libvirtd logs:
Failed to acquire pid file '/run/libvirt/network/driver.pid': Resource temporarily unavailable
I have also seen similar messages referring to librvirtd and qemu*.pid.
I was able to get libvirtd started after explicitly shutting it down and starting it again with systemctl
, but a couple minutes later it shut itself done even as the crc vm kept running just fine. I am pretty sure that libvirtd doesn't actually have anything to do with running the VM, because I can clearly see it isn't running from the Cockpit console.
I wish I understood these subsystems better to be more helpful, but even small data points like this might be helpful.
Recently they split libvirtd into separate services. This is why the main service might be OK in your case, but the network management service was not running (unavailable
). This is specific to Fedora/systemd, and I understand there is an issue and a fix.
We might have to check what we can do to better assist in this case, as we are able to check the state of the services... but not all distros work this way.
$ sudo systemctl enable --now libvirtd.service
I would not recommend this unless you are absolutely sure your setup is not using modular libvirt. If your system is configured for modular libvirt, this could be doing more harm than good.
Fwiw, after installing the libvirt
package on a freshly installed fedora system, all the services below are enabled and running, and virsh -c qemu:///system ...
was functional. libvirtd.service
or libvirtd.socket
are not part of that set.
$ systemctl list-unit-files --all|grep virt |grep enabled
virtlxcd.service enabled enabled
virtqemud.service enabled enabled
virtxend.service enabled enabled
virtinterfaced.socket enabled enabled
virtlockd.socket enabled disabled
virtlogd.socket enabled disabled
virtlxcd-admin.socket enabled disabled
virtlxcd-ro.socket enabled disabled
virtlxcd.socket enabled disabled
virtnetworkd.socket enabled enabled
virtnodedevd.socket enabled enabled
virtnwfilterd.socket enabled enabled
virtproxyd.socket enabled enabled
virtqemud-admin.socket enabled disabled
virtqemud-ro.socket enabled disabled
virtqemud.socket enabled disabled
virtsecretd.socket enabled enabled
virtstoraged.socket enabled enabled
virtxend-admin.socket enabled disabled
virtxend-ro.socket enabled disabled
virtxend.socket enabled disabled
Bit of a follow up. So after a few weeks of updating Fedora 38, I'm not having problems starting crc anymore (at least over a few system reboots for testing purposes), but the VM is completely shut down on every reboot now rather than just being put to sleep. I went into Cockpit and saw it wasn't configured for an automatic start and see if that works, but it's a bit out of scope for this issue. I am still on the penultimate release of crc (2.22.1+e8068b4) that I was on when I originally posted.
Unless more knowledgeable people than me (i.e. almost everyone) can figure out whether anymore time needs to spent on this, seems to me this one doesn't need to be an issue anymore?
autostart will not work, as when the VM is running, we considered CRC started... but in that case the openshift hasnt started as that is something we do as part of the crc start flow
From: hippyod @.>
Sent: Thursday, July 20, 2023 3:31:42 AM
To: crc-org/crc @.>
Cc: Gerard Braad @.>; Comment @.>
Subject: Re: [crc-org/crc] [BUG] [Fedora-38] crc setup
freezes at "Checking if libvirt 'crc' network is available" (Issue #3719)
Bit of a follow up. So after a few weeks of updating Fedora 38, I'm not having problems starting crc anymore (at least over a few system reboots for testing purposes), but the VM is completely shut down on every reboot now rather than just being put to sleep. I went into Cockpit and saw it wasn't configured for an automatic start and see if that works, but it's a bit out of scope for this issue. I am still on the penultimate release of crc (2.22.1+e8068b4) that I was on when I originally posted.
Unless more knowledgeable people than me (i.e. almost everyone) can figure out whether anymore time needs to spent on this, seems to me this one doesn't need to be an issue anymore?
— Reply to this email directly, view it on GitHubhttps://github.com/crc-org/crc/issues/3719#issuecomment-1642644441, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AAAAOZQMUYUXUD3KHB7XEOTXRAYZ5ANCNFSM6AAAAAAZH62RAU. You are receiving this because you commented.Message ID: @.***>
Updated version of systemd now available in f38 which does resolve this issue. I updated my f38 system today and it fixes it. I am closing the issue now.
Just FYI, now the CentOS 9 Stream got same issue. Hope they will resolve that soon.
General information
crc setup
before starting it? I triedCRC version
CRC status
CRC config
Host Operating System
Steps to reproduce
crc setup
crc start
crc stop
crc delete
in preparation to switch to regular OpenShiftcrc setup
Expected
crc setup
exits within a few secondsActual
crc setup
freezes.Logs
I can't run
crc
, and trying to runcrc delete -f
also freezes.