Re: Hope to set up riscv64 ci

[Wei Wu (吴伟) <wuwei2016@iscas.ac.cn>]

Hi,

On 2022/4/15 20:05, Antonio Terceiro wrote:
> Hi,
>
> On Fri, Apr 15, 2022 at 08:59:14AM +0800, Wei Wu (吴伟) wrote:
> > > > The rvlab will support reboot those machine physically. Reboot machine in
> > > > UTC/GMT+08:00(work day) is a good choice.
> > >
> > > I'm curious, how is this going to work? Is there a ticket service? Do we
> > > send e-mail? Weekdays only I assume?
> >
> > Currently there is no specific ticket service or system for it. Just open a
> > issue in [1] and our colleague will restart the machine for you. Or, the
> > more easier way is to just send an email to Bo Yu, and he will take care of
> > the rest communation.
> >
> > It is a known bug(or feature?) in the kernel and/or SBI for Unmatched, and
> > it has been fixed by the latest kernel.
> >
> > [1] https://github.com/plctlab/riscv-lab-access/issues
>
> Is there a ethernet connected power relay, or would you be rebooting the
> boards by hand? If the former, then would it be possible for us to have
> direct out of band access to the boards, i.e. serial console and power
> control?
>
> Maybe from that Jump Host on the Alibaba cloud.

Currently we're rebooting most of the boards in RISC-V Lab (short for 
rvlab) by hand. A few power relay are deploying internally and not 
exposed to the outside world. We're still working on how to safely 
expose these APIs. I appreciated it if who can share some case study 
with us.

Accessing serial console (IIUC the /dev/ttyUSB*) is possible and 
practical. Another distro community asked me to connect the serial 
console to the adjacent boards, e.g. Unmatched031's serial connect to 
032, and vice versa. Just ping me or Bo Yu when it is needed :)


--
Wei Wu (吴伟)
PLCT Lab