-
Task
-
Resolution: Unresolved
-
Highest
-
None
-
None
-
None
-
None
-
The vpp instances are running in virtual environment as part of NSM (networkservicemesh.io). The vpp is started in a privileged kubernetes pod. The pod has access to main network namespace. The interface in question is also virtualized and provided by ovs to the worker node.
uname -r
5.3.18-150300.59.93-default
The vpp instances are running in virtual environment as part of NSM (networkservicemesh.io). The vpp is started in a privileged kubernetes pod. The pod has access to main network namespace. The interface in question is also virtualized and provided by ovs to the worker node. uname -r 5.3.18-150300.59.93-default
Running vpp built from source (main branch) on SLES15 SP4 x86_64 system.
Multiple vpp instances are running on SLES environments with the same configuration and network topology. Some of them stop receiving packets on a host interface (AF_PACKET).
> date; vppctl show interface host-ccd_int; sleep2; date; vppctl show interface host-ccd_int Thu 06 Jul 2023 01:35:27 PM CEST Name Idx State MTU (L3/IP4/IP6/MPLS) Counter Count host-ccd_int 1 up 9000/9000/9000/9000 Thu 06 Jul 2023 01:35:29 PM CEST Defaulted container "forwarder-vpp" out of: forwarder-vpp, forwarder-vpp-init (init) Name Idx State MTU (L3/IP4/IP6/MPLS) Counter Count host-ccd_int 1 up 9000/9000/9000/9000 > ip a s dev ccd_int 20: ccd_int: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc noqueue state UNKNOWN group default qlen 1000 link/ether 86:26:39:3c:16:4b brd ff:ff:ff:ff:ff:ff inet 10.0.40.115/24 brd 10.0.40.255 scope global noprefixroute ccd_int valid_lft forever preferred_lft forever inet6 fe80::8426:39ff:fe3c:164b/64 scope link noprefixroute valid_lft forever preferred_lft forever RX: bytes packets errors dropped missed mcast 4900991634817 22559096386 0 0 0 0 TX: bytes packets errors dropped carrier collsns 4802673411966 21078712696 0 0 0 0 > ss -nep0l Netid State Recv-Q Send-Q Local Address:Port Peer Address:Port p_raw UNCONN 0 0 *:ccd_int * users:(("vpp_main",pid=64820,fd=13)) uid:174088 ino:290992307 sk:4b773 ver:1 cpy_thresh:0 flags( running loss ) ring_rx(blk_size:10485760,blk_nr:1,frm_size:10240,frm_nr:1024,tmo:0,features:0x0) ring_tx(blk_size:10485760,blk_nr:1,frm_size:10240,frm_nr:1024,tmo:0,features:0x0)
The investigation was supported by sykazmi, details can be found in this mail thread
After enabling core dump the SIGABRT signal was sent to vpp. Maybe some further details can be found based on core dump:
$ docker run -ti registry.nordix.org/cloud-native/nsm/vpp-dbg-wcore:main /bin/bash
# gdb --core=/tmp/core/core.vpp_main-1688644886