Re: ruby-devise tmpfail issue on riscv64 debci
Hi,
On Mon, Mar 27, 2023 at 1:43 AM Paul Gevers <elbrus@debian.org> wrote:
>
> Hi Bo,
>
> On 26-03-2023 16:38, Bo YU wrote:
> > BTW, another interesting thing is that if running autopkgtest with
> > `--shell` to interact with it then
> > exit it directly without doing anything, the autopkgtest can remove
> > the testbed successfully.
>
> My *guess* (really, only a wild guess) is that there's a race condition
> here and that the cache is still being written on riscv64 when the
> testbed is destroyed (and the rm of the directory fails because new
> files appeared in it) while on other architectures the writing is done
> before. Your observation matches that wild guess.
Great! You are right.
I just postpone the deletion of testbed with `sleep 10` in d/test/smoke-test.
It works!
```
+ sleep 10
autopkgtest [15:57:28]: test smoke-test: -----------------------]
smoke-test PASS
autopkgtest [15:57:28]: test smoke-test: - - - - - - - - - - results
- - - - - - - - - -
autopkgtest [15:57:29]: @@@@@@@@@@@@@@@@@@@@ summary
smoke-test PASS
vimer@unmatched10:~/debci/ruby-devise/ruby-devise-4.8.1/ruby-devise-4.8.1$
uname -a
Linux unmatched10 6.0.0-3-riscv64 #1 SMP Debian 6.0.7-1 (2022-11-05)
riscv64 GNU/Linux
```
I think if want to see why there's a race condition on riscv64 need
some time. As a workaround,
send a patch to ruby-devise maybe is ok also?
BR,
Bo
>
> Paul
Reply to: