Thanks! We can continue the testing now. :) --Sheng
On Wed, Apr 10, 2013 at 12:28 PM, Marcus Sorensen <shadow...@gmail.com>wrote: > commit f66b9b570f2acb35acfda2d159dcde6fa62390d5 > Author: Marcus Sorensen <mar...@betterservers.com> > Date: Wed Apr 10 13:27:10 2013 -0600 > > Send only \n rather than \r\n to agent socket when sending cmdline > to system VMS > > BUG-ID: CLOUDSTACK-1732 > Signed-off-by: Marcus Sorensen <mar...@betterservers.com> 1365622030 > -0600 > > > > On Wed, Apr 10, 2013 at 1:18 PM, Marcus Sorensen <shadow...@gmail.com>wrote: > >> yes, you're right, I just ran into this today. The \r\n I believed was >> necessary for the socket to flush, but apparently it isn't. If the patch is >> in a request you can apply it. >> >> >> On Wed, Apr 10, 2013 at 12:44 PM, Sheng Yang <sh...@yasker.org> wrote: >> >>> Hi Marcus, >>> >>> I found this 4.2 blocker bug >>> https://issues.apache.org/jira/browse/CLOUDSTACK-1732 caused by your >>> commit to change the communication mechanism for KVM systemvm. >>> >>> As I said, I am not very familiar with python, but seems "\r\n" would >>> generated "^M" rather than normal unix return character. I've tried >>> following fix and it works. But I am not sure if it's the right fix, >>> because obviously "\r\n" is more complex than "\n", so you should had >>> reason to do so. >>> >>> yasker@yasker-devbox:~/develop/cloudstack-oss$ git diff >>> diff --git >>> a/scripts/vm/hypervisor/kvm/patchviasocket.plb/scripts/vm/hypervisor/kvm/ >>> patchviasocket.pl >>> index 443d6e4..7bcd245 100644 >>> --- a/scripts/vm/hypervisor/kvm/patchviasocket.pl >>> +++ b/scripts/vm/hypervisor/kvm/patchviasocket.pl >>> @@ -53,6 +53,6 @@ my $msg = "pubkey:" . $key . "\ncmdline:" . $cmdline; >>> >>> my $socket = IO::Socket::UNIX->new(Peer=>$sockfile,Type=>SOCK_STREAM) >>> or die "ERROR: unable to connect to $sockfile - $^E\n"; >>> -print $socket "$msg\r\n"; >>> +print $socket "$msg\n"; >>> close $socket; >>> >>> So could you shed some lights on it? >>> >>> Thanks! >>> >>> --Sheng >>> >> >> >
