squire_linux hangs and fails to launch radvd

[ChitrangSrivastava]

Prerequisites

• I have read the Contributing Guidelines.
• I have read the Code of Conduct.

Category

Routing (ex. Open/R, VPP, BGP)

Priority

High

Description

In Kernel Layer 3 mode, when CPE_Prefix and CPE_IFACE is set.
squire_linux doesnt run to create radvd file and neither launch radvd daemon. We have made sure that link is up , lo address is assigned and cpe iface is up.
On debugging it is found that Squirelinux hangs in constructor of NetlinkSocket.cpp at this function
possibly in getAllLinks function

Steps to reproduce

1. Setup node in kernel mode with L3
2. observe that radvd is not running and also radvd.conf is not present
3. squire_linux/current also doesnt show any actionable prints

Additional information (logs, etc.)

it seems squire_linux can be restart using sv , could it be because of SIGINT /SIGQUIT/SIGTERM
Have a to restart by killing the squire_linux
Could those handlers need not be registered now with new way of handling ? I remember similar fix was done earlier in platform_linux

Your environment

• M80:

External tickets

No response

[MatthewWEdwards]

envParams.CPE_INTERFACE and envParams.CPE_IFACE_IP_PREFIX are both deprecated fields in node configuration. Are you using these fields? You should use the cpeConfig schema instead. See the schema metadata: https://github.com/terragraph/meta-terragraph/blob/main/src/terragraph-e2e/e2e/config/config_metadata.json#L3046

[ChitrangSrivastava]

Thanks Matthew , I will check that but,
issue is not happening there but it get stuck in squire_linux NetLinkSocket ctor in getAllLinks() function

[ChitrangSrivastava]

Atleeast the script /usr/sbin/config_get_cpe does support reading of old config
Anywhere else to check for ?

and thats the reason squire_linux runs with correct params
/usr/sbin/squire_linux -v 1 -logtostderr --prefix=3021::/64 --nic=nic1 --wireless=