SSH is not working about dhyve

[nlf]

@stongo has been experiencing a similar problem, we're still trying to figure out why

[evanmcc]

me also

[evanmcc]

It looks like /etc/resolv.conf isn't getting generated correctly:

# cat /etc/resolv.conf 
nameserver 192.168.64.1 # eth0
# ifconfig 
docker0   Link encap:Ethernet  HWaddr 02:42:20:53:D7:58  
          inet addr:172.17.42.1  Bcast:0.0.0.0  Mask:255.255.0.0
          UP BROADCAST MULTICAST  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:0 (0.0 B)  TX bytes:0 (0.0 B)

eth0      Link encap:Ethernet  HWaddr DE:F4:CF:47:CF:32  
          inet addr:192.168.64.3  Bcast:192.168.64.255  Mask:255.255.255.0
          inet6 addr: fe80::dcf4:cfff:fe47:cf32/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:44 errors:0 dropped:0 overruns:0 frame:0
          TX packets:38 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:9043 (8.8 KiB)  TX bytes:4168 (4.0 KiB)

However, editing it and saving it (either to 192.168.64.3 or something like 8.8.8.8, which seems to work better from the prompt) doesn't seem to save the image.

When the image is being run directly with xhyve and the resolv.conf ip is 8.8.8.8 I can docker pull stuff, but i still can't ssh in. docker commands that require external access don't work with either 192.169.64.1 or .3.

[evanmcc]

Additionally, the dropbear config(?) is a dead link (to /var/run/dropbear), with no config apparent elsewhere on the file system. If you make that directory, you get to the password prompt, but you can't log in, because /home/docker/.ssh/ is also not present, and I don't know the password. If you change the password, it works. Maybe something is busted in the build script for the dhyveos image?

[nlf]

wait.. so your /etc/dropbear directory is a link to /var/run/dropbear? that's very definitely wrong. password for the docker user is docker. it's very strange that the .ssh directory isn't present.. i'll check it out further and see if i can figure out what's going on.

[evanmcc]

# ls -l /etc/dropbear 
lrwxrwxrwx    1 root     root            19 Aug 24 21:45 /etc/dropbear -> ../var/run/dropbear
# ls /var/run/
acpid.pid        docker.pid       klogd.pid        syslogd.pid
acpid.socket     docker.sock      network          udhcpc.eth0.pid
crond.pid        dropbear.pid     rpcbind.lock     utmp
docker           ifstate          rpcbind.sock
# ls -la /home/docker/
total 0
drwxr-xr-x    2 docker   docker          40 Aug 24 22:00 .
drwxr-xr-x    4 root     root            80 Aug 24 22:00 ..
#

[evanmcc]

fwiw:

md5 bzImage rootfs.cpio.xz
MD5 (bzImage) = eadc6ad46d54d1f335a18c96a39d32b2
MD5 (rootfs.cpio.xz) = 97ccb2bf02eafa405fe851de84fe414c

version is 1.3.0. I meant to try out v1.2.0, but I haven't had the time yet.

[marcbachmann]

I just checked that. ssh is working in v1.2.0

[nlf]

Interesting. I'll check it out and see if I can figure out what happened between the versions

[tt]

I have the same problem. I fixed it by running mkdir /var/run/dropbear.

In the image, /etc/init.d/S50dropbear contains the condition "$(readlink -f /etc/dropbear)" = "/var/run/dropbear". However, /etc/dropbear points to ../var/run/dropbear.

The upstream bug was introduced in buildroot@e7d04dd and fixed in buildroot@29a0f05

It seems the bug was never part of an official release so basing the build on those rather than on snapshots would probably help.

[fab-soundcloud]

I am not so experienced with this. Is there a workaround with the current version of dhyve? If yes, how can I apply it? What are the steps. Thanks!

[nlf]

I'll upload an experimental build right now that you can test to make sure I fixed it correctly

[nlf]

Ok, just uploaded dhyve-os version 1.3.1. If you build a new dhyve vm it should pull that latest version. Please report back your results.

[marcbachmann]

Thanks. It's working now. Great work.

[nlf]

@evanmcc does this new build of dhyve-os resolve your issue too?

[evanmcc]

Kinda. If I manually launch the VM I can ssh in and use docker stuff, but dhyve up hangs forever after booting the VM. I don't have the time to look into it just now, but I'm satisfied that my other issues aren't related to this one.

[evanmcc]

the backticks in this line https://github.com/nlf/dhyve/blob/master/dhyve#L313 cause dhyve up and dhyve down to be executed. This is what's causing the hang I mentioned earlier. Also restarting the VM as instructed does nothing to fix the issue. I'll dig into why set_exports is failing sometime next week.

[fkoehler]

I did a

dhyve down
dhyve destroy

Then I did a dhyve init followed by a dhyve up

The dhyve up shows:

[dhyve] starting....done

but it never exists. It is sort of hanging there.

When I do dhyve status in another terminal window the vm switches between "stopped" and "runnging" state. I can ssh into it, but only if it is in running state. After a short while the connection to the vm is broken (maybe because it is stopped) and it is broken.

I am confused

[fkoehler]

Also when I ssh into it (after it has pulled the 1.3.1 image on init) it still says:

DhyveOS version 1.3.0, build master : a7197ee

[fkoehler]

Oh, I see. When I CTRL+C out of the "[dhyve] starting....done" it keeps running and it seems to stay up normally, meaning I can use it regularly and I can ssh into it.

[marcbachmann]

Works with the newest dhyve-os version.

[marcbachmann]

Please create an issue for the other things, if you still have a problem.