I'm quite new to VMWare Server. All seemed well in testing, was connecting to https://localhost:,RequestProcessor] Error processing action request /view/destroy : No WebAccess to VMware Server 2: Proxysvc, SSL Exception, session id context uninitialized Please log in.
But they don't match with trying to access https.
Does anyone know how to debug this and be able to use the https port again?
Hi,
Well... something is somehow wrong with the certificate...
FWIW I have the same issue here on my local centOS server and haven't been able to fix it so far.
I also performed all of the same steps as you and it happened after upgrading to VMware Server 2.0.2 and changing over to the latest x64 centos.plus kernel.
It worked fine for me with 2.0.1 and the standard kernel using centOS 5.4... So not entirely sure what to blame here yet.
FWIW, I can just connect over the https interface from IE8 on an external machine, then it just works.
But I can't do this from firefox installed locally... so maybe it is even a firefox issue... for the moment I don't have the time to troubleshoot it further, but maybe the extra info helps you in troubleshooting this.
--
Wil
_____________________________________________________
VI-Toolkit & scripts wiki at http://www.vi-toolkit.com
Hi Wila, thanks for the info. You sound like an experience VMWareServer v2 user?
IE6 on XP also can't connect to https - VM host spews same SSL Exception errors - but ffx 3.5.4 on Mac OS X 10.5.8 can, and ffx 3.0.15 on SL4.5 desktop.
But ffx 3.0.15 on my SL5.3 32-bit laptop cannot connect, same SSL Exception errors
On the VM host (also ffx 3.0.15) I tried moving the vmware-admin account's .mozilla directory aside & having ffx make completely fresh .mozilla, but same errors.
So far VMWare is fairly disappointingly fragile. The documentation is not useful for debugging.
/etc/vmware/ssl/rui.{cert,key} are timestamped = when VMware-server-2.0.0-2-3138.x86_64 was installed, but https access broke before that.
Other SSL Exception postings have said things like uninstall vmware server & reinstall - might have to try- seems incredible for so-called production-level software.
Before that, will shutdown VM host & reboot again. Dear me Linux admin should never have to be like this, with decently robust software.....
winniel wrote:Hi Wila, thanks for the info. You sound like an experience VMWareServer v2 user?
Hehe, yeah I gues I'm supposed to be one of those people.
So far VMWare is fairly disappointingly fragile. The documentation is not useful for debugging.
It has always worked perfectly for me before. For me the problem is new and I've never seen it before.
/etc/vmware/ssl/rui.{cert,key} are timestamped = when VMware-server-2.0.0-2-3138.x86_64 was installed, but https access broke before that.
Well that's the thing that surprised me as well as I had not chosen in the setup to regenerate the certificates. Another I just tried was to delete ALL of the certificates I had for this host in my browser and imported therse once more. Hoping that one of the old certificates was causing trouble.
No dice.. I might regenerate the certificate as I have a feeling that something is fishy with that bit...
Other SSL Exception postings have said things like uninstall vmware server & reinstall - might have to try- seems incredible for so-called production-level software.
Before that, will shutdown VM host & reboot again. Dear me Linux admin should never have to be like this, with decently robust software.....
Guess that depends on the linux distribution that you use as there are some distro's out there that require a lot of tinkering. But I agree on a production machine with a production OS it must just work. The tinkering is for at home where it doesn't matter much when things break.
But I also have to admit that the VM's run fine and I can connect to them with VIX and the local console just fine. (See also: Web Access for other ways to access your VM's and control them)
--
Wil
_____________________________________________________
VI-Toolkit & scripts wiki at http://www.vi-toolkit.com
me too! I upgraded CentOS and then installed VMware server 2. vmware-config.pl works fine but when I run vmware
Firefox pops up with https://127.0.0.1:8333/ui/
but then I get a blank screen.
way down in /var/log/vmware/hostd.log there is
Proxysvc SSL Handshake on client connection failed: SSL Exception:
(the square brackets seem to generate an overstrike character)
What version of CentOS?
A clever colleague here running CentOS 5.4 has said he knows of glibc
compatibility problems btw CentOS 5.4 & vmware server v2:
http://bugs.centos.org/view.php?id=3884
http://communities.vmware.com/message/1356119#1356119
He noted that "vmware-hostd doesn't want all the old libraries from the
old glibc package, just libc.so.6"
But later he said it still doesn't fix the problem, still can't connect
to https on his CentOS 5.4 vm host.
My server is ScientificLinux 5.3 but my clever colleague figures out that
the people who support SL backported some glibc fix that essentially
breaks https access to vmware server 2. I don't fully grok why it
only broke https access to vmware server 2 after first reboot, but my
clever colleague seems to grok it.
Firefox pops up with https://127.0.0.1:8333/ui/
but then I get a blank screen.
I see that too, & if hit refresh, firefox change to
Connection Interrupted
The document contains no data.
The network link was interrupted while negotiating a connection. Please try again
While the hostd.log shows about 30 errors all the same:
2009-12-01 10:58:05.497 'Proxysvc' 1081878848 warning SSL Handshake on client connection failed: SSL Exception: error:140D9115:SSL routines:SSL_GET_PREV_SESSION:session id context uninitialized
My conclusion is that vmware server is fragile.
Earlier I said
IE6 on XP also can't connect to https - VM host spews same SSL Exception errors - but ffx 3.5.4 on Mac OS X 10.5.8 can,
and ffx 3.0.15 on SL4.5 desktop.
These are still true, I think adds to the RHEL5/CentOS5/SL5 glibc = likely culprit.
thanks winniel - Winnie L? "My conclusion is that vmware server is fragile. " yeah me too. One option is to install MS Windows on another box.
CentOS 5.4
PBrimacombe said
>One option is to install MS Windows on another box.
The VMs are Scientific Linux as well. The reason we're doing them as VMs is to reduce number of physical servers, since the services they provide aren't advised to be installed on the same machine.
This thread can be left open a bit more in case someone really finds a fix but it looks like glibc in CentOS 5.4 & backported to SL5.3 is the most likely culprit.
I was getting this, too. I just fixed it, but I don't know how.
I'm running VMware Server 2.0.2-203138 on CentOS 5.4. I first had the documented problem with glibc and worked around the problem using the technique described in http://communities.vmware.com/message/1364852.
But I still had the SSL Exception problem. I noticed this work around and logged in as root from a Windows XP machine. Once that worked I created a user ID for myself, and everything worked. Just for grins I tried restarting Firefox on CentOS and tried logging in from CentOS. Magic! Everything works now.
I'd suggest you to try this workaround: https://www.centos.org/modules/newbb/viewtopic.php?topic_id=23555
Hope this helps.
---
Paul Svirin
StarWind Software developer ( http://www.starwindsoftware.com )
Well that worked really well... I tried deleting all firefox vmware.com certificate exceptions (there were 6) from Firefox on Centos. I'm back to where I was earlier this morning. I tried getting on from Windows XP and it looks like the CentOS Firefox problem.
Here's the log. I tried restarting vmware-mgmt, too.
Harry
I started receiving this same error today on a CentOS 5.4 x86_64 system. I had been using the libc.so.6 workaround until this morning when I ran an update. On loading the http://helios:8222 web page the browser screen blanks and remains "loading". A look at the /var/log/vmware/hostd.log shows lots of SSL errors.
I tried deleting all the certificates from my browser. On reload I had to create exceptions for the sites, but same issue.
I tried re-installing the VMWare-server package with rpm -ivh --replacepkgs VMware-server-xxxx.rpm but same error (had to redo the glibc workaround is this will overwrite the vmware-hostd script that's edited).
I tried manually killing the hostd process, clearing the /var/run lockfiles, then restarting. This failed. So I tried running vmware-hostd from the command line. It initially gave an error about finding the libvmomi.so.1.0 library so I added /usr/lib/vmware/vmacore to LD_LIBRARY_PATH then re-ran ldconfig. That got past the missing library error but an strace on the command shows lots of EBADF errors..Not sure if that's even relevant, however. No joy.
I know that the hostd process is listening as I can attach to it with strace and see activity whenever I try to connect via the browser.
I backed out the kernel to the previous version and re-ran the vmware-config.pl.
I've even gone to the point of trying a different java binary to replace the jre1.5.0_15, but no luck.
At this point I'm considering blowing away the installation and trying ESXi, but that means I lose the functionality of the host OS. I'm considering VirtualBox and Xen/KVM, but would really like to see this working. Anyone have any luck getting this to work?
Hi,
I have the same problem, but was pretty convinced nowadays that the issue is about the combination of firefox and vmware Server 2 not working well on a centOS 5.4 x64 setup. At least not when using the secure web interface.
So... I was looking for a way to test that theory.. I wanted to use epiphany, but it doesn't appear to be in the repository, so I tried konqueror, but got javascripts errors after logging in (so much that you get in a forever debug loop, hurray!)
Then it finally dawned on me to use the opera browser.
Downloading and installing opera took me about 1 minute and voila, it works great!
So umm... I think the problem is firefox related in one way or the other...
--
Wil
_____________________________________________________
VI-Toolkit & scripts wiki at http://www.vi-toolkit.com
edit: OK scratch that "great" part as Opera has its own issues, but those are not related to the secure web interface as the same issues appear when using the plain HTTP interface. I now also blogged about this issue here: VMware Server 2, Proxysvc SSL Handshake problems
shecki wrote:We try to solve a problem with FF3.6 and VMserver2 in the inofficially german VMware-community too.
First point on the road was to reenable SSL2-protocol in the about:config dialog.
>
>securitiy.enable.ssl2 = false -> true
Now the user have access to the web-interface but the console is dead. Normally on a click inside the black console-window with the play-icon a new window should open but all you get is a timeout. If you make a shortcut from the vm on the desktop then you get console access...