Friday, March 27, 2009

EON 64-bit 0.59.0 based on SNV_110 is released!

Embedded Operating system/Networking (EON), RAM based live ZFS NAS appliance is released on Genunix! Much thanks to for download hosting and serving the opensolaris community.

It is available in a CIFS and Samba flavor
tryitEON 64-bit x86 CIFS ISO image version 0.59.0 based on snv_110
tryitEON 64-bit x86 Samba ISO image version 0.59.0 based on snv_110

tryitEON 32-bit x86 CIFS ISO image version 0.59.0 based on snv_110

tryitEON 32-bit x86 Samba ISO image version 0.59.0 based on snv_110New/Fix:
- should work with vdi/vmware disks properly.
- image footprint/runtime is smaller requires less RAM.


David said...

Dear Andre,

I've been looking around for some time for a ZFS NAS distro and this is the first one besides FreeNAS to offer these two features.
Unfortunately, I'm having trouble understanding how this thing works (being a Solaris newcomer, this may be understandable) so I was wondering if there was some kind of documentation available that explains the basic workings of this distro. Also, I was wondering how to add stuff to it, say some kind of fancy WebGUI (FreeNAS *hint*) as the groundwork for that (Apache/PHP) seems to be in place already.

Thanks for this great piece of software!

David said...

Forgot to ask this in the post above: would there be a way to add XFS support to it? I saw that there was stable read-only and experimental write support available in solaris and that would greatly help in migrating from my current linux/XFS to eon/ZFS.

Thanks again.

Andre Lue said...

Hi David,

Please see:

There is a start on Freenas interface copy but it is stalled because of time. You can grab a copy here but its only informational at this stage. It has no control calls.

This is a ZFS only image. Do you need XFS support to just transfer the data from another NAS? Would sftp/scp or nfs mount suffice for now?

Feel free to browse the posts for more info.

David said...


THanks for the quick reply. I missed that first post, which is helpful. Are the buildscripts available somewhere?
Also, I asked about XFS because I would like to transfer the data internally (SATA->SATA) as copying 6TB over sftp/scp or NFS takes very, very long.

Andre Lue said...

Hi David,

Don't think there is XFS support in opensolaris. You would probably have better luck creating your zpool and then using fuse-zfs to copy the data but i'm not sure if would be faster than 1gigE transfer.

Eoin said...

This is a fantastic piece of work and almost exactly what I need for a number of applications.
However, as an OpenSolaris newbie I am struggling a bit with joining an EON (64-bit CIFS 0.59) appliance to a domain. In particular, it doesn't seem to have a DNS client running, which would make it impossible. Or am I missing something?

tchu said...


Great thanks for marvelous job!
EON looks very promising.
And I'm here with commonly asked question - where I can get a build scripts? I would like to create EON image with COMSTAR iscsi target suppport.


Andre Lue said...


For DNS resolution you simply create /etc/resolv.conf

and enter the respect dns resolver info, and using smbadm join should get you what you're after.


The build scripts are here. Note I included native iSCSI not the comstar bits as it's a bit bigger footprint.

Andre Lue said...


You would also have to cp nsswitch.dns to nsswitch.conf in /etc.

Eoin said...
This comment has been removed by the author.
Eoin said...

Hey Andre,

I'm still having trouble joining my fileserver to my domain using EON 0.59 (cifs version, same result with both 64-bit and 32-bit), even though using this hardware joins the same domain with both 2008.11 and 2009.06 (111a) installed from the liveCD.

When I run:
smbadm join -u administrator ecit.local

the response is:
failed to join ecit.local: INTERNAL_ERROR

and /var/svc/log/network-smb-server:default.log shows a core dump:
[ May 11 00:23:18 Stopping because process dumped core. ]
[ May 11 00:23:18 Executing stop method (:kill). ]
[ May 11 00:23:18 Executing start method ("/usr/lib/smbsrv/smbd start"). ]
smbd: NetBIOS services started
[ May 11 00:23:19 Method "start" exited with status 0. ]

Oddly enough, using the right domain name and user name with a bad password gives "failed to find any domain controllers for ecit.local"

Can you suggest anything?

Andre Lue said...


Do a truss and post the output of /tmp/client to

truss -fae -vall -o /tmp/client smbadm join -u administrator ecit.local

Just before doing this also do

truss -fae -vall -o /tmp/svr -p pid_of_smbd_start

snoop -o /tmp/domain host IP_of_domain_controller

This may require alot of memory depending on the capture size. It should not be a problem if you have a zfs swap configured.

Eoin said...

Hope I got this right!
This one is a VMware server 2.0 vm.

Andre Lue said...


Hard to tell without the other 2 logs. VMware + solaris are famous for having clock issues. Can you make sure time between your domain controller and vm is sync'd.

Just to make sure, you have fulfilled all the steps for a domain controller setup?

Did you try the community scripts cifs-chkcfg or cifs-gendiag ?

Eoin said...

Hi Andre,

I take your point about VMware clock issues, but I get the same result on a physical machine. The clock is right, and when I kill xntp, I can run ntpdate successfully against the domain controller. It may help to know that attempting to join a domain gives these two messages at the console:

May 12 11:46:13 eon smbd[208]: [ID 362282 daemon.error] ads: Retry kinit to acquire credential.
May 12 11:46:13 eon smbd[208]: [ID 305693 daemon.error] ads: major status error: An unsupported mechanism was requested

I have followed all the standard procedures for joining other than modifying /etc/krb5/krb5.conf, but I have joined two 2008.11 and a 2009.6 servers to two different Win2003 domains without touching krb5.conf and they all work perfectly.

cifs-chkcfg fails with the message: "line 37: pkginfo: command not found \ some CIFS packages are missing" (naturally enough), but once the package check code is removed, it completes with just one message:
"svcs: Pattern 'samba' doesn't match any instances"

cifs-gendiag generates a lot of info, but I am not able to interpret any of it as a reason why domain joining fails. Given my consistent success with Sun releases and consistent failure with EON 0.59 across different hardware, different domains and both 32 & 64 bit versions, all minimally configured from the distribution as released, I am pretty sure that my results are generally reproducible. Has anyone joined 0.59 to a domain? If so, I'd love to compare notes.

I may try a previous EON release, but it's of little practical use to me, as my core objective is to expose the snapshots through the Windows Explorer 'Previous Versions' tab. That needs build 106+.

Andre Lue said...


Can you try to send me the other truss and snoop when you do the truss of the smbadm?


Andre Lue said...


You are correct. I can reproduce this. I will continue trying to debug this further. Please feel free to share any progress you make on this.


Andre Lue said...

Eoin, All

I am seeing the exact behavior in this thread, so the core dump when joining a domain seems opensolaris related.

I am targeting a snv_114 release as snv_113 has other pausing issues.

Some issues fixed since snv_112:
- multi-core cpu support
- seperated backup file list from
- added marvell88sx and AHCI driver support (intel BADAXE2 now supported)
- dtrace support added

Robert said...

Hi Andre,

sadly I got now stucked at the domain join bug.
I assume the bug will be cleared with SNV_114? Do you have already an idea when the new version will be available?

I'll move away from the current opensolaris installation asap - the idmapd memory leak drives me crazy, but the domain join is essential for me.

BTW: Thanks for this fantastic compilation :-)