-
Bug
-
Resolution: Not a Bug
-
Undefined
-
None
-
rhel-9.5
-
None
-
None
-
None
-
rhel-sst-virtualization-storage
-
ssg_virtualization
-
None
-
False
-
-
None
-
None
-
None
-
None
-
None
Description of problem:
block-threshold event for virtio disk sometimes late for more than 30s
Version-Release number of selected component (if applicable):
libvirt-10.4.0-1.el9.x86_64
How reproducible:
80%
Steps to Reproduce:
1. Start a vm with 2 virtio disks
prepare the test.qcow3 image
# cd /var/lib/libvirt/images # qemu-img create -f qcow2 test.qcow2 200M # virsh dumpxml rhel --xpath //disk <disk type="file" device="disk"> <driver name="qemu" type="qcow2"/> <source file="/var/lib/libvirt/images/RHEL-9.5-x86_64-latest-ovmf.qcow2"/> <target dev="vda" bus="virtio"/> <address type="pci" domain="0x0000" bus="0x03" slot="0x00" function="0x0"/> </disk> <disk type="file" device="disk"> <driver name="qemu" type="qcow2"/> <source file="/var/lib/libvirt/images/test.qcow2"/> <target dev="vdb" bus="virtio"/> <address type="pci" domain="0x0000" bus="0x04" slot="0x00" function="0x0"/> </disk>
2. Set domblkthreshold for vdb:
# virsh domblklist rhel Target Source --------------------------------------------------------------------- vda /var/lib/libvirt/images/RHEL-9.5-x86_64-latest-ovmf.qcow2 vdb /var/lib/libvirt/images/test.qcow2 # virsh domblkthreshold rhel vdb 100M # virsh domstats rhel --backing | grep threshold block.1.threshold=104857600
3. write data above the threshold on the vm, check if there is event on host:
On 1st terminal, track the events:
# virsh event --event block-threshold --loop --timestamp
On 2nd terminal, login vm and write data to vdb:
[guest] # mkfs.ext4 /dev/vdb
[guest] # mount /dev/vdb /mnt
[guest] # dd if=/dev/urandom of=/mnt/bigfile bs=1M count=110 ; date
110+0 records in
110+0 records out
115343360 bytes (115 MB, 110 MiB) copied, 0.458109 s, 252 MB/s
Mon Jun 17 20:31:27 CST 2024
The captured log is about 31s late:
# virsh event --event block-threshold --loop --timestamp 2024-06-17 12:31:58.474+0000: event 'block-threshold' for domain 'rhel': dev: vdb(/var/lib/libvirt/images/test.qcow2) 104857600 524288 2024-06-17 12:31:58.474+0000: event 'block-threshold' for domain 'rhel': dev: vdb[1](/var/lib/libvirt/images/test.qcow2) 104857600 524288
Actual results:
block-threshold event for virtio disk sometimes late for more than 30s
Expected results:
The block-threshold event log immediately after the operation
Additional info:
1) Can also be reproduced on rhel 10 with libvirt-10.4.0-1.el10.x86_64, can not reproduce with sata disk;
2) It seems when using larger numbers (more than the threshold), no such issue:
Terminal 1:
# dd if=/dev/urandom of=/mnt/bigfile bs=1M count=150 ; date
150+0 records in
150+0 records out
157286400 bytes (157 MB, 150 MiB) copied, 0.775307 s, 203 MB/s
Mon Jun 17 20:47:14 CST 2024
Terminal 2:
# virsh domstats rhel --backing | grep threshold block.1.threshold=104857600 # virsh event --event block-threshold --loop --timestamp 2024-06-17 12:47:14.566+0000: event 'block-threshold' for domain 'rhel': dev: vdb(/var/lib/libvirt/images/test.qcow2) 104857600 589824 2024-06-17 12:47:14.566+0000: event 'block-threshold' for domain 'rhel': dev: vdb[1](/var/lib/libvirt/images/test.qcow2) 104857600 589824
