On 9/6/2016 6:03 PM, Michal Privoznik wrote:
On 05.09.2016 18:42, Maxim Nestratov wrote:
> From: Yuri Pudgorodskiy <yur(a)virtuozzo.com>
>
> A separate error code will help recognize real failures from
> necessity to try again
>
> Signed-off-by: Maxim Nestratov <mnestratov(a)virtuozzo.com>
> ---
> include/libvirt/virterror.h | 2 ++
> src/qemu/qemu_agent.c | 6 +++---
> src/util/virerror.c | 6 ++++++
> 3 files changed, 11 insertions(+), 3 deletions(-)
ACK
Just curious - have you experienced this behaviour?
Michal
.
Issue any GA command while domains is just booted gives a good chance
of the following:
- domain is started running
- virtio agent.0 channel is already connected to guest, but qemu-ga is
not ready yet
- a libvirt client issues any GA commands
- command begins with guest-sync written to virtio serial device buffer
- qemu-ga did not respond within default timeout (5 seconds), client got
TIMEOUT
- a client repeated GA commands, waiting for GA to come up, buffer is
populated with repeated guest-sync command
- GA finally reads command from virtio, responding to id from the 1st
guest-sync found in channel (FIFO)
- but libvirt now expects last id, and responds with VIR_ERR_INTERNAL_ERROR
- ... VIR_ERR_INTERNAL_ERROR looks terrible from the client point of view
Moreover, any timeout from qemu-ga results in VIR_ERR_AGENT_UNSYNCED error.
Same scenario:
- two guest agent commands one after another
- 1st timeouted
- libvirt is expecting response with id of the 2nd
- qemu-ga finally comes with answer to the 1st
- libvirt is under if (id_ret != id) condition and is responding with
VIR_ERR_AGENT_UNSYNCED