Closed renzhengeek closed 1 year ago
Is there a way to login guest to debug?
I tried this: /usr/local/bin/socat - VSOCK-CONNECT:3:1025
but got:
2023/06/16 09:23:44 socat[235757] E connect(5, AF=40 cid:3 port:1025, 16): No such device
I'm sure the cloud-hypervisor is running with vsock device cid 3.
/usr/local/bin/cloud-hypervisor --api-socket /run/kuasar-vmm/9115cd3cf5d02518994f1e8af0c5cbe61e07c53b1c7fee69dfd33effe969acdd/api.sock --cpus boot=1,affinity=,features= --memory size=1073741824,shared=on,hugepages=off --kernel /var/lib/kuasar/vmlinux.bin --cmdline console=hvc0 root=/dev/pmem0p1 rootflags=data=ordered,errors=remount-ro ro rootfstype=ext4 task.sharefs_type=virtiofs task.log_level=debug --pmem id=rootfs,file=/var/lib/kuasar/kuasar.img,discard_writes=on --rng src=/dev/urandom --vsock cid=3,socket=/run/kuasar-vmm/9115cd3cf5d02518994f1e8af0c5cbe61e07c53b1c7fee69dfd33effe969acdd/task.vsock,id=vsock --console file=/tmp/9115cd3cf5d02518994f1e8af0c5cbe61e07c53b1c7fee69dfd33effe969acdd-task.log --fs tag=kuasar,socket=/run/kuasar-vmm/9115cd3cf5d02518994f1e8af0c5cbe61e07c53b1c7fee69dfd33effe969acdd/virtiofs.sock,id=fs -v
seems vmm-task is running:
grep -i task /tmp/9115cd3cf5d02518994f1e8af0c5cbe61e07c53b1c7fee69dfd33effe969acdd-task.log
[ 0.090767] registered taskstats version 1
[2023-06-16T01:00:38.032382Z INFO vmm_task] Task server start with config: TaskConfig { sharefs_type: "virtiofs", log_level: "debug" }
[2023-06-16T01:00:38.032490Z DEBUG vmm_task] listen vsock port 1025 for debug console
[2023-06-16T01:00:38.032570Z DEBUG vmm_task::device] no scsi driver installed
[2023-06-16T01:00:38.034318Z INFO vmm_task] Task server successfully started, waiting for exit signal...
...
BTW, the slack link in contact section of readme page, seems not pointing to a kuasar room. Can we add slack service address in README?
@renzhengeek Thaks for report. I believe after #27 merged, this bug can be fixed.
If cloud hypervisor is used as underlay vmm, you should connect the unix socket that CLH has listened, see: https://github.com/cloud-hypervisor/cloud-hypervisor/blob/main/docs/vsock.md#connecting-from-host-to-guest
I will check the slack link, thank you.
Hi @Burning1020 ,
@renzhengeek Thaks for report. I believe after #27 merged, this bug can be fixed.
It works! thansk for the quick fix. 👍
If cloud hypervisor is used as underlay vmm, you should connect the unix socket that CLH has listened, see: https://github.com/cloud-hypervisor/cloud-hypervisor/blob/main/docs/vsock.md#connecting-from-host-to-guest
Thanks for the pointer~
What happened?
Hi developers,
I followd “Quick start” doc: https://github.com/kuasar-io/kuasar#quick-start to play the vmm sandboxer, and got stumbled at the last steps. Please help to debug the issue:-)
crictl -D -r unix:///run/containerd/containerd.sock run --runtime=vmm container.json pod.json failed at starting containers:
vmm-sandboxer output: (not full debug log, only key infos) RUST_LOG="debug" /root/kuasar/vmm/sandbox/target/debug/vmm-sandboxer --listen /run/vmm-sandboxer.sock --dir /run/kuasar-vmm
containerd:(I added some debug log in container create path):
What did you expect to happen?
examples/run_example_container.sh vmm
vmm examples works, container task gets running!
How can we reproduce it (as minimally and precisely as possible)?
follow: https://github.com/kuasar-io/kuasar#quick-start
Anything else we need to know?
No response
Dev environment
No response