Warning: SunRay + VB 3.2.8 + Bridged NIC = reboot

[kebabbert]

I have OpenSolaris b134 and SunRay Server Software. When I had VB 3.1.4 everything worked fine. Now I just recently upgraded to VB 3.2.8 and it is a catastrophe. My computer reboots sometimes. SunRay does not work. Now I have rebooted to a earlier Snapshot with VB 3.1.4 and everything works fine. My conclusion is that VB 3.2.8 messes up the computer horribly in some way. So I advice against upgrading to v3.2.8 if you use SunRay. I dont know how to debug this problem.

Oracle Solaris 11 Express is coming at the end of this year, now that OpenSolaris has officially been killed. So maybe VB team should just ignore this bug report and focus on S11 Express instead?


UPDATE: It seems the problem is a bug in OpenSolaris and not in VirtualBox.

[Ramshankar]

The VirtualBox team focuses on VirtualBox. If a bug, regardless of the host platform, comes to our attention, we do our best to fix it. That being said, please be aware the free, public bug ticket resolution is on a best effort basis.

Now that that's cleared up, the best way to help us fix bugs is by opening a public bug ticket in our wonderful Bugtracker (http://www.virtualbox.org/wiki/Bugtracker). Please make sure you include all relevant information there, including VBox.log, and in this case (any interesting bits in /var/adm/messages). A system that rebooted/panicked would usually give good indications as to why. Please read http://www.virtualbox.org/wiki/Core_dump on how to enable system core dumps for OpenSolaris if you are unfamiliar with it. It could help us immensely in solving such bugs.

Thank you!

[kebabbert]

Ok, I have narrowed down the problem. The problem is that SunRay thin clients does not work anymore. It is as, I have not configured any SunRay clients, they dont show the login screen. Sometimes the computer reboots too. The SunRay clients use an IP adress from the server.

When I use Bridged NIC on my VDI, this problem appears. When I use NAT, there are no problems. It is as if Bridged NIC VDIs interfere with SunRay client IP adresses. But NAT VDIs dont disturb IP adresses to the SunRay.

So workaround is to use NAT, and not Bridged VDIs.

[Ramshankar]

Sometimes the computer reboots too.

That shouldn't happen. Sounds like a bug, could you perhaps check if it left any evidence in the syslog (file /var/adm/messages)?

The SunRay clients use an IP adress from the server.

That's normal with bridged networking. The two modes are quite different and thus you should only use the networking mode you require. The user manual explains them pretty well. But in any case it must not spontaneously reboot the host.

Thank you for the feedback.

[kebabbert]

Regarding reboot and /var/adm/messages. That files is 1.7MB and I dont remember which date or what to look for. If you wish, I can attach it for you.

Regarding SunRay IP. But VB should work in conjunction with SunRay in bridged mode, yes? Now, I must use NAT, otherwise hell breaks loose. Bridged NIC is a no-no

[Ramshankar]

Yes, If you could the "messages" file in some place it might be helpful.

Regarding bridged networking mode, yes you should not have problems like reboots etc., but I don't have enough information regarding how you've configured networking in your systems to know if bridged gives you the desired result in terms of IP assignment etc. NAT and bridged are quite different modes and thus only you should know which would work in your networking setup. My point is, no matter what networking mode - bridged or otherwise, running VirtualBox should not panic the host. Hence my request for the syslog.

[kebabbert]

Ok, yesterday I booted my computer and before login screen appeared, it rebooted. Just as I experienced earlier. The second time it booted fine. I never had this problem with earlier VB, no matter bridged or NAT NIC.

I managed to identify and cut out the part of /var/adm/messages where OpenSolaris b134 rebooted and I attach it here. I hope this is the correct part of messages file. If not, please tell me what more information you need.

[kebabbert]

Hmm... When I try to browse the network in NAT mode, it does not work. I must use bridge mode?
http://forums.virtualbox.org/viewtopic. ... 90&start=0

So... this means I am screwed if I use SunRay and NAT network, because then I can not access the network? Maybe I better downgrade to v3.1.4? Which version introduced this bridged problem? v3.2.8 reboots the computer. v3.1.4 does not reboot. Maybe v3.1.8 is safe?

Do you find anything obvious in the "/var/adm/messages" file that I can try out?

[Ramshankar]

Those crashes don't even seem to be VirtualBox related from the stack crawl in the logs. Is this the only system in which you encounter spontaneous reboots? If these are servers could you check the iLOM/eLOM logs to see if it picked up any hardware errors (like on the FSB)?

[kebabbert]

Those crashes don't even seem to be VirtualBox related from the stack crawl in the logs. Is this the only system in which you encounter spontaneous reboots? If these are servers could you check the iLOM/eLOM logs to see if it picked up any hardware errors (like on the FSB)?

This is just my home PC. Intel Quad 9450.

I had some problems earlier with OpenSolaris though in builds later than 125(?). It was sloooow and sound lagged. It turned out that the power saving mechanism in OpenSolaris was to aggressive, so I turned power saving off and everything worked fine.

So you say the problem does not appear to be VB? What could the problem be? In SunRay software? Do you want me to attach additional log files?

PS. Thanks for your help looking into this problem.

[Ramshankar]

The crash certainly looks strange. It seems to be somewhere in the ProcFS path, but of course can't be sure what is triggering this with just the trace. The reason I asked to check if it could be hardware related was because we've not received any similar reports (and certainly nothing like the stack crawl you've shown).

[jkeil]

I managed to identify and cut out the part of /var/adm/messages where OpenSolaris b134 rebooted and I attach it here.

The panic backtrace looks very similar to

Code: Select all

Bug ID   	 6932278
Synopsis 	panic in procfs:prchoose+72

http://bugs.opensolaris.org/bugdatabase ... id=6932278

It's a duplicate of 6924278 amd might be fixed in build 137.

[kebabbert]

jkeil,
As there are no b137 available, do you have any suggestions on what I can do to solve this problem? Any work arounds? Are there newer builds available somewhere? etc?

So, we can conclude that the problem is not in VirtualBox then?

[Ramshankar]

I think we can safely say it is definitely not a VirtualBox bug. Can't really answer your question on newer builds though as that's the (Open)Solaris area.