Working DSM 6


Recommended Posts

Of course, the boot it's just for boot :smile:

My initial reply on this matter was addressed to a user that allready added a second drive (ignoring the boot drive from count). After that second drive is added in dsm it's safe to remove the first one.

Like I said, personally I see no reason to do that but every one is entitled to their own setup :smile:

 

Sent from my SM-G930F using Tapatalk

Link to post
Share on other sites
Hello

 

I managed to create a DSM6 under ESXi6 (using the OVF), but I am unable to add a drive for the volume.

 

I use the free ESXi hypervisor with the free client.

 

any tips?

 

-Roei

 

make sure you are running ESXi 6U2 - then use the web UI : https://ESXI_HOST/ui/

 

for anyone else using ESXi and want the easy way to update :

http://www.v-front.de/2016/01/how-to-use-esxi-patch-tracker-to-update.html

Link to post
Share on other sites

I was looking around in this DSM 6 a bit. I had an idea, would it not be possible to extract the VM images or something like this from the in beta Virtual DSM and see how they made them work. Cause I see when you make a virtual DSM you also need to go true the install with it's own pat file. I know this version is a little bit more limited then the normal one but for must users it could be sufficient. And it could also be a pretty versatile build, cause it's perfect for virtualisation.

Link to post
Share on other sites
I was looking around in this DSM 6 a bit. I had an idea, would it not be possible to extract the VM images or something like this from the in beta Virtual DSM and see how they made them work. Cause I see when you make a virtual DSM you also need to go true the install with it's own pat file. I know this version is a little bit more limited then the normal one but for must users it could be sufficient. And it could also be a pretty versatile build, cause it's perfect for virtualisation.

 

It's totally possible to run the VirtualDSM version instead of one made for real hardware. That's not the problem. The problem is that we don't have the drivers for any non-Synology hardware, and we can't compile them since Synology hasn't realease the kernel source code, or at least the kernel headers. Without that, no matter which version you want to run, you can't do it on hardware for which the version doesn't have the drivers.

 

cause it's perfect for virtualisation

It's perfect for virtualisation in KVM. VirtualDSM uses VirtIO drivers, which are, AFAIK, only supported by KVM.

Link to post
Share on other sites

Is it possible to use this method for kvm? In the end the vmdk files provided can be converted which I did. The installation went fine but the only drawback is that the "run" kernel doesn't include the network drivers I guess. Because if I boot with "run" option the installed synology can not be found in my network anymore.

Link to post
Share on other sites

Oktisme I know what you mean. My idea was more cause they did use some virtual machine hosts system (kvm in this case as you said), it would maybe be possible to just extract the boot image that they use for it. Cause it's all standard virtual KVM hardware. But if you say it's more based on direct real hardware access we are for sure back to the non released headers problem. It was just an idea to get around this problem in a creative way.

Link to post
Share on other sites
Just an update looking a bit in the Virtualisation package I see it's just QEMU. So nothing fancy I would say.

 

Yeah but VirtualDSM talks with the real DSM to get some info about CPU, serial number, and some other things. You can't just pick it up and put it in KVM somewhere else.

You could emulate the host's side of the communication, or patch VirtualDSM.

Link to post
Share on other sites
Just an update looking a bit in the Virtualisation package I see it's just QEMU. So nothing fancy I would say.

 

Yeah but VirtualDSM talks with the real DSM to get some info about CPU, serial number, and some other things. You can't just pick it up and put it in KVM somewhere else.

You could emulate the host's side of the communication, or patch VirtualDSM.

 

Yeah of course. I just was thinking it's maybe a bit easier with drivers already present optimized for a VM situation. I could then patch VirtualDSM as suggested. I also looked in the DockerDSM package. It has no serialnumber check or something because there is no bootloader, but spotted an extra port connection to verify the host system.

 

So then the simplest way to get it going for now is your approach with the E1000 nic on ESxi 6.

Link to post
Share on other sites
It has no serialnumber check or something because there is no bootloader

 

Even on the real DSM there's no protection in the boot loader, so DockerDSM not having a boot loader doesn't mean anything.

 

So then the simplest way to get it going for now is your approach with the E1000 nic on ESxi 6.

 

Well, to get it working right now, yeah.

If you really want it running in QEMU, there's this thread on the QEMU mailing list about adding e1000e emulation. I havent checked it all so I don't know if it got merged or not.

Link to post
Share on other sites

I know it's a little bit offtopic, but the docker implementation in dsm6 realy became nice :smile:

 

docker-compose is preinstalled (in 5.2 you had to install it yourself) and can be used from the shell to provision composed docker containers to run a single application - the installed containers are integrated nicely in the Docker UI and can be used the same way like single containers would be used (same is true for 5.2 if you install docer-compse and import containers with).

 

docker networking is supported by the ui and allows to use bridget and host mode. The host mode should allow containers to have their own ip :wink:

The host mode has two advantages: less overhead (huge advantage for download client like sabnzbd or nzbget) and you get a direct ip connection (if you don't force the trafic over a reverse-proxy). The drawbacks: you have to take care of what you do, since port mapping does not applay anymore.

 

Some Containers seem to need less RAM in DSM6: the gitlab container itself needed 1.2gb in DSM5.2, while it needs roughly less then 700mb ram in 6.0 (I reused the datavolume from 5.2 in 6.0 without further changes and used the same container versions)

 

It is a pitty that DockerDSM does not work.. (pat file can be downloaded at http://download.synology.com/download/D ... M_7393.pat, DSM_DockerDSM_7321.pat does not work either, because the DockerDSM version must match the DSM version)

Last line in dmesg is: [81486.608219] synoddsm-ctnd[20380]: segfault at 7f7c9760102c ip 00007f732634ceea sp 00007ffefd385d20 error 4 in ld-2.20-2014.11.so[7f7326344000+21000]

Link to post
Share on other sites

I know it's a little bit offtopic, but the docker implementation in dsm6 realy became nice :smile:

docker-compose is preinstalled (in 5.2 you had to install it yourself) and can be used from the shell to provision composed docker containers to run a single application - the installed containers are integrated nicely in the Docker UI and can be used the same way like single containers would be used (same is true for 5.2 if you install docer-compse and import containers with).

docker networking is supported by the ui and allows to use bridget and host mode. The host mode should allow containers to have their own ip :wink:

The host mode has two advantages: less overhead (huge advantage for download client like sabnzbd or nzbget) and you get a direct ip connection (if you don't force the trafic over a reverse-proxy). The drawbacks: you have to take care of what you do, since port mapping does not applay anymore.

Some Containers seem to need less RAM in DSM6: the gitlab container itself needed 1.2gb in DSM5.2, while it needs roughly less then 700mb ram in 6.0 (I reused the datavolume from 5.2 in 6.0 without further changes and used the same container versions)

It is a pitty that DockerDSM does not work.. (pat file can be downloaded at http://download.synology.com/download/D ... M_7393.pat, DSM_DockerDSM_7321.pat does not work either, because the DockerDSM version must match the DSM version)

Last line in dmesg is: [81486.608219] synoddsm-ctnd[20380]: segfault at 7f7c9760102c ip 00007f732634ceea sp 00007ffefd385d20 error 4 in ld-2.20-2014.11.so[7f7326344000 21000]

Maybe DockerDSM has to be patched too?

 

 

Posted via Xpenology.us

Link to post
Share on other sites

For what it's worth... you only get a single DockerDSM license. This might be suitable for people that are running it on bare metal and want to use the DockerDSM for all services that are exposed to the internet...

 

Actually I got DockerDSM up and running now: install DockerDSM, but don't let it start automaticly. Open the details tab and use the start button from there.

Otherwise permissions inside DockerDSM become corrupted: no execution right on almost everything in /usr/bin and /usr/sbin.

Though: it might be that the permissions are fine because I deleted DockerDSM, removed the data folder (/volume1/DockerDSM1) and reinstalled it.

 

Not sure, either one of both is responsible that the permissions don't get spoiled.

 

Another oddity:

if i use a static ip, the host is not accessible from outside. From DockerDSM you can ping DSM6, but nothing else.

If you use dhcp, no network interface is available at all.

 

Then i tried another approach: i imported the tar file that was inside the pat file with docker. It become available in the list of images and allowed to start a container. Though, it seem that the container needs some envoirenment variables and volumes in order to work properly. I couldn' t figure out yet what DockerDSM passes over to the container in order to make it work. If the variables and volumes are identified, we should be able to start the container and use the DockerDSM-Image withouth using the DockerDSM function itself. Also i am sure that this approach would bypass the license check *cough*

Link to post
Share on other sites

Then i tried another approach: i imported the tar file that was inside the pat file with docker. It become available in the list of images and allowed to start a container. Though, it seem that the container needs some envoirenment variables and volumes in order to work properly. I couldn' t figure out yet what DockerDSM passes over to the container in order to make it work.

Can you share the tar file? If we get docker DSM working in any host that would be nice :razz:

 

 

 

Posted via Xpenology.us

Link to post
Share on other sites
Then i tried another approach: i imported the tar file that was inside the pat file with docker. It become available in the list of images and allowed to start a container. Though, it seem that the container needs some envoirenment variables and volumes in order to work properly. I couldn' t figure out yet what DockerDSM passes over to the container in order to make it work. If the variables and volumes are identified, we should be able to start the container and use the DockerDSM-Image withouth using the DockerDSM function itself. Also i am sure that this approach would bypass the license check *cough*

 

This is what I was looking into trying last night - my ultimate aim is to be able to run DSM inside a Docker container on my Ubuntu host, at the moment I'm using a VM and a boot script to replace volume1 with a NFS mount (yes it actually works somehow).

 

I managed to inspect the container created by DockerDSM last night and get the volume mounts and environment variables, but I gave up after I couldn't get DockerDSM itself working, so I thought it would be pointless trying to get it working on another host. However now you guys have got it working on DSM, I might give this a go again and feed back some info here.

 

I took the Docker Image that gets built by the Docker package and pushed it up to my own registry, I could push that to Docker Hub which might be a good start.

Link to post
Share on other sites

Would you mind sharing enviroment and volume mounts?

 

Is there any particular reason why you used nfs to mount volume1? I doubt that DSM uses NFS to mount a share inside the container. That's what Docker volumes are for :smile:

 

DockerDSM maps the host share /volume1/DockerDSM1/data to /volume1 in the container.

But tDockerDSM1 contains another folder with configuration files, e.g. with network setting. i'd like to understand where those get mounted :wink:

 

Maybee it's time for an own thread?

Link to post
Share on other sites
Would you mind sharing enviroment and volume mounts?

 

Is there any particular reason why you used nfs to mount volume1? I doubt that DSM uses NFS to mount a share inside the container. That's what Docker volumes are for :smile:

 

DockerDSM maps the host share /volume1/DockerDSM1/data to /volume1 in the container.

But tDockerDSM1 contains another folder with configuration files, e.g. with network setting. i'd like to understand where those get mounted :wink:

 

Maybee it's time for an own thread?

 

Sorry I meant I use NFS to mount volume1 on my current Xpenology project, not DockerDSM. It's because I moved away from Synology Hybrid Raid to my own BTRFS RAID running on an Ubuntu Server, but I still wanted to use the Synology DSM frontend and packages, so I just mount a folder on my BTRFS filesystem into the Xpenology VM and I'm good to go :grin:

 

I'll have a play around with this again today or tomorrow and get back to you :smile:

Link to post
Share on other sites

I got DSM 6 working on a PC, not VM.

 

The only way I got it:

 

1. Restore default configuration, in a installed 5.2-5644 PC system.

2. Boot with Xpenoboot 5.2-5644 USB pendrive

3. Access web for first config, but not use "manual installation", I choose "install now".

4. After several minutes downloading from synology, the system reboots with actual DSM 6.0.1-7393.

 

I don't know if people can't update their systems or the problems about upgrading are VMs (VMware, ESXi) related.

For now, the only service that doesn't work is "Resource Monitor", if open a message about network error appears, and close automatically.

 

Thanks for the Xpenology community and developers!!!

xpe.PNG.3fea90bbaf8c930f531cefcd53eeb17e.PNG

Link to post
Share on other sites
Guest
This topic is now closed to further replies.