Bug 1272311
Summary: | Without vhost=on, during netperf w/ protocol UDP_STREAM in guest, this guest can not response other thing | ||
---|---|---|---|
Product: | Red Hat Enterprise Linux 7 | Reporter: | weliao <weliao> |
Component: | qemu-kvm-rhev | Assignee: | jason wang <jasowang> |
Status: | CLOSED DUPLICATE | QA Contact: | Virtualization Bugs <virt-bugs> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 7.2 | CC: | ailan, chayang, hhuang, huding, jasowang, juzhang, knoel, michen, mst, rbalakri, stefanha, virt-bugs, virt-maint, weliao, xfu |
Target Milestone: | rc | ||
Target Release: | --- | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | 962733 | Environment: | |
Last Closed: | 2016-07-04 02:40:39 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: | |||
Bug Depends On: | 962733, 1272312 | ||
Bug Blocks: | 962749 |
Comment 2
weliao
2015-10-16 04:33:48 UTC
What's the qdisc used in host? Need to use something which can have fair queueing. E.g sfq or others. Suspect a duplication of https://bugzilla.redhat.com/show_bug.cgi?id=955540. A known issue. The solution is using enabling tx interrupt in guest. (In reply to jason wang from comment #3) > What's the qdisc used in host? Need to use something which can have fair > queueing. E.g sfq or others. > > Suspect a duplication of https://bugzilla.redhat.com/show_bug.cgi?id=955540. > A known issue. The solution is using enabling tx interrupt in guest. Ok, speak too fast. You see the issue if vhost=off? Need your virtio-net part of info qtree. Does it work if you're using tx=timer? (In reply to jason wang from comment #3) > What's the qdisc used in host? Need to use something which can have fair > queueing. E.g sfq or others. > > Suspect a duplication of https://bugzilla.redhat.com/show_bug.cgi?id=955540. > A known issue. The solution is using enabling tx interrupt in guest. [root@dhcp-8-118 home]# ip link show 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 2: eno1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master ovs-system state UP mode DEFAULT qlen 1000 link/ether 24:be:05:0c:1e:92 brd ff:ff:ff:ff:ff:ff 3: ovs-system: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT link/ether 82:5c:83:46:ae:3c brd ff:ff:ff:ff:ff:ff 4: pri-ovs1: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT link/ether 4a:4a:a9:97:b5:4a brd ff:ff:ff:ff:ff:ff 5: ovs0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN mode DEFAULT link/ether 24:be:05:0c:1e:92 brd ff:ff:ff:ff:ff:ff 6: privatebr: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT link/ether fe:b5:f8:03:87:48 brd ff:ff:ff:ff:ff:ff 7: pri-ovs0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT link/ether 1a:71:96:6e:fa:42 brd ff:ff:ff:ff:ff:ff 8: virbr0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN mode DEFAULT link/ether 52:54:00:e3:d7:db brd ff:ff:ff:ff:ff:ff 9: virbr0-nic: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast master virbr0 state DOWN mode DEFAULT qlen 500 link/ether 52:54:00:e3:d7:db brd ff:ff:ff:ff:ff:ff 24: tap0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master ovs-system state UNKNOWN mode DEFAULT qlen 500 link/ether 0e:13:04:7c:40:66 brd ff:ff:ff:ff:ff:ff Ok, speak too fast. You see the issue if vhost=off? Need your virtio-net part of info qtree. Does it work if you're using tx=timer? dev: virtio-net-pci, id "net0" ioeventfd = false vectors = 3 (0x3) virtio-pci-bus-master-bug-migration = false disable-legacy = false disable-modern = true addr = 03.0 romfile = "pxe-virtio.rom" rombar = 1 (0x1) multifunction = false command_serr_enable = true class Ethernet controller, addr 00:03.0, pci id 1af4:1000 (sub 1af4:0001) bar 0: i/o at 0xffffffffffffffff [0x1e] bar 1: mem at 0xffffffffffffffff [0xffe] bar 6: mem at 0xffffffffffffffff [0x3fffe] bus: virtio-bus type virtio-pci-bus dev: virtio-net-device, id "" csum = true guest_csum = true gso = true guest_tso4 = true guest_tso6 = true guest_ecn = true guest_ufo = true guest_announce = true host_tso4 = true host_tso6 = true host_ecn = true host_ufo = true mrg_rxbuf = true status = true ctrl_vq = true ctrl_rx = true ctrl_vlan = true ctrl_rx_extra = true ctrl_mac_addr = true ctrl_guest_offloads = true mq = false mac = "52:54:00:0b:06:81" vlan = <null> netdev = "hostnet0" x-txtimer = 150000 (0x249f0) x-txburst = 256 (0x100) tx = "" indirect_desc = true event_idx = true notify_on_empty = true any_layout = true So you've already used tx timer? How about tx bh? Is this a regression? If yes, can you bisect to find the first bad commit. If not, can you reproduce this issue on upstream qemu.git? Thanks (In reply to jason wang from comment #6) > So you've already used tx timer? How about tx bh? Is this a regression? If > yes, can you bisect to find the first bad commit. If not, can you reproduce > this issue on upstream qemu.git? > > Thanks if tx = "bh", has this issue too, retest on qemu-kvm-rhev-2.3.0-2.el7.x86_64, still has this issue,So it is not a regression bz. (In reply to weliao from comment #7) > (In reply to jason wang from comment #6) > > So you've already used tx timer? How about tx bh? Is this a regression? If > > yes, can you bisect to find the first bad commit. If not, can you reproduce > > this issue on upstream qemu.git? > > > > Thanks > > if tx = "bh", has this issue too, > retest on qemu-kvm-rhev-2.3.0-2.el7.x86_64, still has this issue,So it is > not a regression bz. The question was answered partially. Can you reproduce this on upstream qemu.git? Could not reproduce the issue. Needs: - tc qdisc show in both host and guest - brctl show in host - use simple cli, e.g just keep network and disk and remove all other devices to see it the problem still. Thanks Host: [root@dhcp-8-139 home]# tc qdisc show qdisc pfifo_fast 0: dev eno1 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: dev virbr0-nic root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: dev tap0 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 [root@dhcp-8-118 qemu-kvm]# ovs-vsctl show d888d2ed-3336-4429-98f7-0b922604cb70 Bridge "ovs0" Port "ovs0" Interface "ovs0" type: internal Port "tap0" Interface "tap0" Port "eno1" Interface "eno1" ovs_version: "2.4.0" Guest: [root@dhcp-9-25 home]# tc qdisc show qdisc pfifo_fast 0: dev eth0 root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 qdisc pfifo_fast 0: dev virbr0-nic root refcnt 2 bands 3 priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 (In reply to jason wang from comment #8) > (In reply to weliao from comment #7) > > (In reply to jason wang from comment #6) > > > So you've already used tx timer? How about tx bh? Is this a regression? If > > > yes, can you bisect to find the first bad commit. If not, can you reproduce > > > this issue on upstream qemu.git? > > > > > > Thanks > > > > if tx = "bh", has this issue too, > > retest on qemu-kvm-rhev-2.3.0-2.el7.x86_64, still has this issue,So it is > > not a regression bz. > > The question was answered partially. Can you reproduce this on upstream > qemu.git? qemu-system-x86_64 -name guest1 -S -machine pc,accel=kvm,usb=off -cpu SandyBridge -m 4G -smp 4 -drive file=/home/rhel7.2.qcow2,if=none,id=drive-virtio-disk0,format=qcow2 -device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x7,drive=drive-virtio-disk0,id=virtio-disk0 -boot menu=on -netdev tap,script=/etc/ovs-ifup,downscript=/etc/ovs-ifdown,id=hostnet0,vhost=off -device virtio-net-pci,netdev=hostnet0,id=net0,mac=52:54:00:0b:06:81,bus=pci.0,addr=0x3 -monitor stdio -qmp tcp:0:4444,server,nowait -vnc 0.0.0.0:1 -L /usr/share/seabios/ -bios bios-256k.bin yes, can reproduce this issue on upstream qemu.git. *** This bug has been marked as a duplicate of bug 1293233 *** *** This bug has been marked as a duplicate of bug 954261 *** |