On Tue, Aug 16, 2011 at 02:52:07PM -0400, Dave Allan wrote: > On Tue, Aug 16, 2011 at 11:15:04AM -0700, Daniel P. Berrange wrote: > > On Mon, Aug 15, 2011 at 02:19:27PM -0400, Dave Allan wrote: > > > On Mon, Aug 15, 2011 at 10:12:54AM -0700, Daniel P. Berrange wrote: > > > > From: "Daniel P. Berrange" <berrange@xxxxxxxxxx> > > > > > > > > After running 'virsh console' in interactive mode, there was a > > > > missing call to virStreamAbort, which meant the server kept the > > > > stream resources open > > > > > > The combination of this patch and the other stream related patch makes > > > the problem I was seeing of garbled console following multiple console > > > connect/disconnects go away, but another problem remains in this area. > > > After connecting to the console in interactive virsh and > > > disconnecting, I often see all subsequent commands fail: > > > > > > virsh # dominfo foo > > > error: failed to get domain 'foo' > > > error: An error occurred, but the cause is unknown > > > > > > virsh # dominfo foo > > > error: failed to get domain 'foo' > > > error: no call waiting for reply with prog 536903814 vers 1 serial 300 > > > > > > virsh # define bar.xml > > > error: Failed to define domain from bar.xml > > > error: no call waiting for reply with prog 536903814 vers 1 serial 301 > > > > > > Closing and reopening the interactive virsh session makes the problem > > > go away. > > > > > > Powering down the guest while the console is connected has made the > > > problem 100% reproducible in the 4 or 5 tries I've given it. > > > > Yeah this is a regression. We were accidentally sending back two > > stream abort RPC packets in some scenarios (with race condition). > > > > I've posted a patch to fix it > > It does indeed fix the failure on domain shutdown, but there is still > a problem related to high volumes of console output. If I get on the > console and execute: > > while true ; do echo -n "The quick brown fox jumped over the lazy > dogs. The quick brown fox jumped over the lazy dogs. "; done > > and exit the console while the output is spraying, I still see a > failure: > > the lazy dogs. The quick brown fox jumped over the lazy dogs. The > quick brown fox jumped over t > virsh # console foo > error: failed to get domain 'foo' > error: An error occurred, but the cause is unknown > > virsh # console foo > error: failed to get domain 'foo' > error: An error occurred, but the cause is unknown > > virsh # console foo > error: failed to get domain 'foo' > error: no call waiting for reply with prog 536903814 vers 1 serial 6 > > virsh # > > > BTW, I noticed this first when exiting the console while boot messages > were being output; I'm not *trying* to break it, and many thanks for > looking into all this. :) This time i found a bug that pre-dated my RPC refactoring :-) Sent 2 more patches.... Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :| -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list