Closed ghost closed 9 years ago
The VirtualBox logs tells something about failed to start the UDP tunnel server:
00:00:00.132793 UDPTunnel#0: sport=10004;dest=192.168.1.10;dport=10005
00:00:00.132933 UDPTunnel#1: sport=10006;dest=192.168.1.10;dport=10007
00:00:00.133053 UDPTunnel#2: sport=10008;dest=192.168.1.10;dport=10009
00:00:00.133170 UDPTunnel#3: sport=10010;dest=192.168.1.10;dport=10011
00:00:00.133197 VMSetError: /Users/vbox/tinderbox/4.3-mac-rel/src/VBox/Devices/Network/DrvUDPTunnel.cpp(556) int drvUDPTunnelConstruct(PDMDRVINS*, CFGMNODE*, uint32_t); rc=VERR_PDM_HIF_OPEN_FAILED
00:00:00.133211 VMSetError: UDPTunnel: Failed to start the UDP tunnel server
00:00:00.133236 VMSetError: /Users/vbox/tinderbox/4.3-mac-rel/src/VBox/Devices/Network/DevE1000.cpp(7709) int e1kR3Construct(PDMDEVINS*, int, CFGMNODE*); rc=VERR_PDM_HIF_OPEN_FAILED
00:00:00.133236 VMSetError: Failed to attach the network LUN
00:00:00.133236 PDM: Failed to construct 'e1000'/4! VERR_PDM_HIF_OPEN_FAILED (-2852) - Unable to open the host interface.
00:00:00.137635 ERROR [COM]: aRC=NS_ERROR_FAILURE (0x80004005) aIID={8ab7c520-2442-4b66-8d74-4ff1e195d2b6} aComponent={Console} aText={UDPTunnel: Failed to start the UDP tunnel server (VERR_PDM_HIF_OPEN_FAILED).
00:00:00.137646 Failed to attach the network LUN (VERR_PDM_HIF_OPEN_FAILED)}, preserve=false
00:00:00.139414 Power up failed (vrc=VERR_PDM_HIF_OPEN_FAILED, rc=NS_ERROR_FAILURE (0X80004005))
In https://community.gns3.com/thread/8516 the same VirtualBox error was shown, when he starts using more than 4 connections. I don't know if both issues are connected.
Update 1: They seem to be connected !!! If I remove PC4 I have no problems starting both VirtualBox VMs at once.
Update 2: User ngs3 confirmed, that he has the same problem on Windows 7 (64 bit)
I don't have the same issue with MacOS 10.10 and VirtualBox 4.3.26 I try to upgrade VirtualBox
Idem with 4.3.28 and the following topology
Work also for this:
You have to configure more links. My topology works, if I remove PC4. My current assumption is, that you need at least 2 virtual box VMs with 4 active interfaces each.
And I had no problems with this topology:
So you need something similar to my topology
Your vEOS are VirtualBoxes ? And the PC4 is a VPCS?
Yes, the vEOS are VirtualBox VMs. Currently I use the linked clone feature, but for the test I created two full clones and used them.
The PC1 - PC4 are VPCS.
In https://community.gns3.com/message/27540#27540 the topology is attached (untitled.gns3)
This work for me:
I will look deeply in your topology
This crash for me:
Could you test with: http://sourceforge.net/projects/gns-3/files/Nightly%20Builds/2015-06-01/GNS3-stable.2015-06-01.dmg/download
It should start vm one by one
No, that doesn't work, it looks like before. Well, not really exactly. Neither with 1.3.3 nor with the test version I could reproduce the error with 4 VPCS, I increased them to 6 VPCS. But then the error was exactly the same with both versions.
But I'm not sure, I have your changed version.
GNS3.app is actual:
/Users/behlers# ls -l /Volumes/GNS3/
total 8
lrwxr-xr-x 1 behlers staff 13 1 Jun 14:17 Applications@ -> /Applications
drwxr-xr-x 3 behlers staff 102 1 Jun 14:16 GNS3.app/
But the contents of GNS3.app/Contents/Resources/lib/python34.zip is from today 05:16, but that may be because the time zone is not in the zip file.
/Users/behlers# unzip -l /Volumes/GNS3/GNS3.app/Contents/Resources/lib/python34.zip
Archive: /Volumes/GNS3/GNS3.app/Contents/Resources/lib/python34.zip
Length Date Time Name
-------- ---- ---- ----
0 06-01-15 05:16 ./
4138 06-01-15 05:16 __future__.pyo
983 06-01-15 05:16 _bootlocale.pyo
23916 06-01-15 05:16 _collections_abc.pyo
4310 06-01-15 05:16 _compat_pickle.pyo
4792 06-01-15 05:16 _dummy_thread.pyo
10619 06-01-15 05:16 _osx_support.pyo
15585 06-01-15 05:16 _strptime.pyo
...
Of course I didn't start the GNS3 from the dmg, I copied it to /Applications.
You can check the version in the about dialog or in the console at the bottom.
I have the feeling Vbox OSX is not very reliable with large number of adapter...
Jeremy do you have the same issue on Linux?
Sure, i'm not jeremy :-), is he following this issue ?
Anyway I created this whole issue, because someone had problems with virtualbox on windows 7, see https://community.gns3.com/thread/8516. So don't think it's OS X specific.
Yeah jeremy watch everything :)
I just tried on my debian 8 (Jessie) with gns3 v1.3.3.
I copied the vdi file from OS X and created 2 vEOS vbox VMs, the same as in OS X. I then created a topology with 2 vEOS VMs and 12 VPCS ( 6 per vEOS). No problems.
But there are some differences:
I understand the issue :) It's related to VPCS.
VPCS allocate an UDP port at source + 9 sometimes it's collide with VirtualBox
In GNS3 VPCS is started using -i1 which should prevent that to happen or maybe there is a problem there?
The options is here, perhaps vpcs ignore it
I confirm same issue with vpcs trunk version
The issue is here: https://sourceforge.net/p/vpcs/code/HEAD/tree/trunk/src/relay.c#l318
I found a way to dirty patch it by turning off the relay mode.
Fixed by a VPCS patch
GNS3 V1.3.3, OS Mac OS X V10.9.5, VirtualBox 4.3.28
See https://community.gns3.com/message/27540#27540
Topology:
If I start all devices at once, one of the VirtualBox VMs doesn't start. It's reproducible, it happens all the time. They are not linked clones, just 2 "normal" VMs. When I start one VM, wait a few seconds and then start the rest, everything works fine.
The log shows: