QUORUM_REPORT_BAD events are limited to a maximum rate of 1 per second. While this is not a problem in itself, this means that an error in one a Quorum child will mask errors in the other children if they happen within the same 1 second interval.
This series fixes that problem by separating these events in different queues if they come from different nodes. Once we add the 'type' field to QUORUM_REPORT_BAD we will also be able to classify them according to the type if we want. In addition to the above, this series also fixes a crash that happens if there's an I/O error in one of the children. This is serious enough so I'll send the patch to fix this crash to qemu-stable as well. Regards, Berto v2: - Patch 3: Rename clock_type to event_clock_type [Max] v1: https://lists.gnu.org/archive/html/qemu-devel/2016-03/msg02161.html - Original version Alberto Garcia (4): quorum: Fix crash in quorum_aio_cb() monitor: Separate QUORUM_REPORT_BAD events according to the node name monitor: Use QEMU_CLOCK_VIRTUAL for the event queue in qtest mode iotests: Add test for QMP event rates block/quorum.c | 12 +++-- monitor.c | 22 ++++++-- tests/qemu-iotests/146 | 129 +++++++++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/146.out | 5 ++ tests/qemu-iotests/group | 1 + 5 files changed, 161 insertions(+), 8 deletions(-) create mode 100644 tests/qemu-iotests/146 create mode 100644 tests/qemu-iotests/146.out -- 2.7.0
