It was me. The reason is explained in the patch description (you can find more details in description of the commit, introduced the hook). I'm waiting for khorenko@ to convince him, that the idea to "podkovat' blohy" was wrong by design and we should drop the hook. Once I succeed, I'll sent appropriate series to get rid of it completely. So this patch is temporary one, because too many tests have failed because of this race.
2 окт. 2017 г. 8:05 PM пользователь Andrey Vagin <ava...@virtuozzo.com> написал: On Mon, Oct 02, 2017 at 06:47:19PM +0400, Stanislav Kinsburskiy wrote: > This hook is needed only for releaseing venet IP address early (thus allowing > to restart container with the same IP faster). > But mount are destroyed asynchroniosly, and thus NFS mount can be destroyed > after IP address is dropped. > Let's fir this race it the way how all the world does things: release IP with > network namespace. Why don't you remove this code? Who added this hook? What was a reason? > > https://jira.sw.ru/browse/PSBM-73193 > > Signed-off-by: Stanislav Kinsburskiy <skinsbur...@virtuozzo.com> > --- > kernel/ve/ve.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/kernel/ve/ve.c b/kernel/ve/ve.c > index b0188c3..4efb9d9 100644 > --- a/kernel/ve/ve.c > +++ b/kernel/ve/ve.c > @@ -451,7 +451,8 @@ static void ve_drop_context(struct ve_struct *ve) > synchronize_rcu(); > put_nsproxy(ve_ns); > > - ve_hook_iterate_fini(VE_SHUTDOWN_CHAIN, ve); > + /* This have to be revisited */ > +// ve_hook_iterate_fini(VE_SHUTDOWN_CHAIN, ve); > > put_cred(ve->init_cred); > ve->init_cred = NULL; > > _______________________________________________ > Devel mailing list > Devel@openvz.org > https://lists.openvz.org/mailman/listinfo/devel
_______________________________________________ Devel mailing list Devel@openvz.org https://lists.openvz.org/mailman/listinfo/devel