VirtualBox

Opened 6 years ago

Closed 6 years ago

#17352 closed defect (duplicate)

NAT network gateway becomes unresponsive

Reported by: datadestroyd Owned by:
Component: network/NAT Version: VirtualBox 5.2.0
Keywords: Cc:
Guest type: Linux Host type: Windows

Description

I have an issue with the NAT network component. I have multiple Oracle Linux 6.5 x64 guests set up with a NAT network 192.168.123.0/24 (on a Windows 7 x64 host). No DHCP, static IP addresses only. The network works just fine at first, but then suddenly the gateway becomes completely unresponsive: no traffic gets out, ping 192.168.123.1 returns destination host unreachable. This affects all machines in this network. The only way to get it working again is to power down all VMs and restart VirtualBox.

The weird thing is, this always happens during a specific unit test of the OCI8 extension for PHP, which I was "make test"ing at the time. It's the test "imp_res_cancel.phpt" from OCI8 version 2.0.12. I'm using the latest Oracle Instant Client. The tests are configured to use a database outside the virtual network. So something this test is sending over the network must be disrupting the virtual NAT.

If anyone has any idea how I can investigate this further, please let me know.

Attachments (2)

FEMU.log (1007 bytes ) - added by datadestroyd 6 years ago.
tcp_dump_1521_vm.txt.xz (481.4 KB ) - added by datadestroyd 6 years ago.
Last 50000 lines only

Download all attachments as: .zip

Change History (9)

comment:1 by Valery Ushakov, 6 years ago

Please can you attach .VirtualBox/networkname.log

Please can you also capture the traffic during the offending test? Both on the host (on the interface the test peer is on) and on the guest where the other test peer is on.

Last edited 6 years ago by Valery Ushakov (previous) (diff)

comment:2 by Valery Ushakov, 6 years ago

Also you can restart the natnet with

VBoxManage natnetwork stop --netname NatNetwork && \
VBoxManage natnetwork start --netname NatNetwork

by datadestroyd, 6 years ago

Attachment: FEMU.log added

comment:3 by datadestroyd, 6 years ago

I attached the network log. However, nothing is written to the log when the error occurs. I also tried to capture the traffic inside the VM using tcpdump, but when the error occurs, tcpdump just hangs indefinitely (but according to ps still consumes about 3% CPU). I'll try to capture the traffic on the host next.

One more thing: When I run the aforementioned test alone, the error does not occur. So it appears to be a problem that builds up over time.

Last edited 6 years ago by datadestroyd (previous) (diff)

by datadestroyd, 6 years ago

Attachment: tcp_dump_1521_vm.txt.xz added

Last 50000 lines only

comment:4 by Valery Ushakov, 6 years ago

Please, attach real traffic capture (a pcap file), not text output.

comment:5 by datadestroyd, 6 years ago

pcap file as requested (from the VM) can be downloaded here: https://gigamove.rz.rwth-aachen.de/d/id/Pg5kHYP8YHf785

I can't attach it to the ticket due to its size.

comment:7 by Valery Ushakov, 6 years ago

Resolution: duplicate
Status: newclosed
Note: See TracTickets for help on using tickets.

© 2023 Oracle
ContactPrivacy policyTerms of Use