About the function of setup_host.sh.

[hongyi-zhao]

Hi,

I try to figure out the function of the setup_host.sh script but still not so clear.

Any hints will be highly appreciated.

Best regards, HY

[daohu527]

The comment has been explained very clearly. Generally speaking, it's do some configurations.

1. Setup core dump format. when the program crash, there will be a coredump file used to record the error and debug the problem.
2. Setup ntpdate to run once per minute. Log at /var/log/syslog. For time synchronization
3. Add udev rules. udev allows for rules that specify what name is given to a device, regardless of which port it is plugged into
4. Add uvcvideo clock config. Set usb camera clock config

[storypku]

The comment has been explained very clearly. Generally speaking, it's do some configurations.

1. Setup core dump format. when the program crash, there will be a coredump file used to record the error and debug the problem.
2. Setup ntpdate to run once per minute. Log at /var/log/syslog. For time synchronization
3. Add udev rules. udev allows for rules that specify what name is given to a device, regardless of which port it is plugged into
4. Add uvcvideo clock config. Set usb camera clock config

One question about "Enable Core Dump on Host": Since apollo was run inside docker, why do we need to enable coredump on the host machine where docker is running on?

I was wondering the reason why the original author wrote it, but got no clue ...

Do you have any ideas ? @xiaoxq @Capri2014 @jinghaomiao @ycool

[daohu527]

Seems modify the config in dockerfile will have some problem, and It is best to modify outside docker. ref link link

First make sure the container will write the cores to an existing location in the container filesystem. The core generation settings are set in the host, not in the container. Example:

echo '/cores/core.%e.%p' | sudo tee /proc/sys/kernel/core_pattern

[hongyi-zhao]

2. Setup ntpdate to run once per minute. Log at /var/log/syslog. For time synchronization

For the real car running on the load, say, the Apollo D-Kit system, IMO, the time is obtained and/or synced from/among the GPS/GNSS/RTK/Lidar systems, etc..

So, I'm still not so clear why should we use ntpdate for this job and do it so frequently.

Best regards, HY

[Capri2014]

Because Apollo modules need precise time stamps to collaborate and perform especially in complex environments, and system time may have accumulated error and differ from GPS time. You would find this frequent time sync essential also after your initial hundreds of hours road tests.

[hongyi-zhao]

@Capri2014 OK, thanks a lot for your explanations. It seems the job in this script only need to be run once immediately after the system has been started or the user has logged into. So, if I understand correct, we only need to apply a check workflow, say, in /etc/profile or ~/.profile, for confirmation whether the correct settings have been applied via invoking this script. Am I right?

Best regards, HY

[storypku]

Especially, do we need to run it as a very beginning step each time when I want to start apollo?

@hongyi-zhao I don't think setup_host.sh need to be run each time you started Apollo. You only need to run it once on the host.

[hongyi-zhao]

Thanks. Got it.