[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 17:12
Message generated for change (Settings changed) made by jessorensen
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
>Status: Closed
>Resolution: Out of Date
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

>Comment By: Jes Sorensen (jessorensen)
Date: 2010-08-30 08:57

Message:
Andy,

Thanks for the reply, I'll close this as out of date. If someone else hits
this problem they
can go ahead and open a new bug in launchpad.

Cheers,
Jes


----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2010-08-27 23:16

Message:
Well, I haven't used rtl8139 for a long time, so I almost forgot about this
problem.

I think the report can be closed, as there are plenty of alternatives
which will work for the average user (including virtio which I use now). If
somebody cares explicitly about rtl8139, he can reopen this bug later.

----------------------------------------------------------------------

Comment By: Jes Sorensen (jessorensen)
Date: 2010-08-19 12:22

Message:
Hi,

Trying to catch up on old bugs and clean up the queue.

Are you still seeing this problem? If not would you mind closing the bug?

For even more performance, you should try out virtio-net.

Cheers,
Jes


----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-08-09 10:54

Message:
Logged In: YES 
user_id=1815
Originator: YES

Indeed, setting model=pcnet also works (same guest OS - Ubuntu 8.04). But
still I like e1000 more now :-))


----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 08:04

Message:
Logged In: YES 
user_id=2174347
Originator: NO

It *seems* that a 'pcnet'-type doesn't shutdown the network emulation, can
you
repro? (I seem to have issues under Plan 9 with the e1000 nic).


Sincerely,
Jerome

----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 07:51

Message:
Logged In: YES 
user_id=2174347
Originator: NO

I've observed the same behavior with kvm-72 and 2.6.26.2 on intel,
the guest is a (very) up to date plan 9 operating system.

I have tried ne2k_pci AND rtl8139 emulation with tap, still, after
performing some file transfers I seem to lose network connectivity.

Please note that I do not need to stress the network card significantly
to lose network connectivity, transferring a file of roughly 20MB
appears to be sufficient.

I haven't tried a e1000 gigE nic at this time, let me know if other
options or workaround are available?


Thanks,
Jerome

----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-06-29 18:09

Message:
Logged In: YES 
user_id=1815
Originator: YES

Update: I was advised on irc to try e1000 network card emulation instead,
and with it I get very good results - 32 megabytes per second and more. So
the bug occurs only with rtl8139 emulation, perhaps because of bandwidth
limitation (packets aren't read quick enough from the TAP handle)?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 19:12
Message generated for change (Comment added) made by andyz
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2010-08-28 01:16

Message:
Well, I haven't used rtl8139 for a long time, so I almost forgot about this
problem.

I think the report can be closed, as there are plenty of alternatives
which will work for the average user (including virtio which I use now). If
somebody cares explicitly about rtl8139, he can reopen this bug later.

----------------------------------------------------------------------

Comment By: Jes Sorensen (jessorensen)
Date: 2010-08-19 14:22

Message:
Hi,

Trying to catch up on old bugs and clean up the queue.

Are you still seeing this problem? If not would you mind closing the bug?

For even more performance, you should try out virtio-net.

Cheers,
Jes


----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-08-09 12:54

Message:
Logged In: YES 
user_id=1815
Originator: YES

Indeed, setting model=pcnet also works (same guest OS - Ubuntu 8.04). But
still I like e1000 more now :-))


----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 10:04

Message:
Logged In: YES 
user_id=2174347
Originator: NO

It *seems* that a 'pcnet'-type doesn't shutdown the network emulation, can
you
repro? (I seem to have issues under Plan 9 with the e1000 nic).


Sincerely,
Jerome

----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 09:51

Message:
Logged In: YES 
user_id=2174347
Originator: NO

I've observed the same behavior with kvm-72 and 2.6.26.2 on intel,
the guest is a (very) up to date plan 9 operating system.

I have tried ne2k_pci AND rtl8139 emulation with tap, still, after
performing some file transfers I seem to lose network connectivity.

Please note that I do not need to stress the network card significantly
to lose network connectivity, transferring a file of roughly 20MB
appears to be sufficient.

I haven't tried a e1000 gigE nic at this time, let me know if other
options or workaround are available?


Thanks,
Jerome

----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-06-29 20:09

Message:
Logged In: YES 
user_id=1815
Originator: YES

Update: I was advised on irc to try e1000 network card emulation instead,
and with it I get very good results - 32 megabytes per second and more. So
the bug occurs only with rtl8139 emulation, perhaps because of bandwidth
limitation (packets aren't read quick enough from the TAP handle)?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 17:12
Message generated for change (Comment added) made by jessorensen
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

>Comment By: Jes Sorensen (jessorensen)
Date: 2010-08-19 12:22

Message:
Hi,

Trying to catch up on old bugs and clean up the queue.

Are you still seeing this problem? If not would you mind closing the bug?

For even more performance, you should try out virtio-net.

Cheers,
Jes


----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-08-09 10:54

Message:
Logged In: YES 
user_id=1815
Originator: YES

Indeed, setting model=pcnet also works (same guest OS - Ubuntu 8.04). But
still I like e1000 more now :-))


----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 08:04

Message:
Logged In: YES 
user_id=2174347
Originator: NO

It *seems* that a 'pcnet'-type doesn't shutdown the network emulation, can
you
repro? (I seem to have issues under Plan 9 with the e1000 nic).


Sincerely,
Jerome

----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 07:51

Message:
Logged In: YES 
user_id=2174347
Originator: NO

I've observed the same behavior with kvm-72 and 2.6.26.2 on intel,
the guest is a (very) up to date plan 9 operating system.

I have tried ne2k_pci AND rtl8139 emulation with tap, still, after
performing some file transfers I seem to lose network connectivity.

Please note that I do not need to stress the network card significantly
to lose network connectivity, transferring a file of roughly 20MB
appears to be sufficient.

I haven't tried a e1000 gigE nic at this time, let me know if other
options or workaround are available?


Thanks,
Jerome

----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-06-29 18:09

Message:
Logged In: YES 
user_id=1815
Originator: YES

Update: I was advised on irc to try e1000 network card emulation instead,
and with it I get very good results - 32 megabytes per second and more. So
the bug occurs only with rtl8139 emulation, perhaps because of bandwidth
limitation (packets aren't read quick enough from the TAP handle)?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 19:12
Message generated for change (Comment added) made by andyz
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

>Comment By: Andrew Zabolotny (andyz)
Date: 2008-08-09 12:54

Message:
Logged In: YES 
user_id=1815
Originator: YES

Indeed, setting model=pcnet also works (same guest OS - Ubuntu 8.04). But
still I like e1000 more now :-))


----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 10:04

Message:
Logged In: YES 
user_id=2174347
Originator: NO

It *seems* that a 'pcnet'-type doesn't shutdown the network emulation, can
you
repro? (I seem to have issues under Plan 9 with the e1000 nic).


Sincerely,
Jerome

----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 09:51

Message:
Logged In: YES 
user_id=2174347
Originator: NO

I've observed the same behavior with kvm-72 and 2.6.26.2 on intel,
the guest is a (very) up to date plan 9 operating system.

I have tried ne2k_pci AND rtl8139 emulation with tap, still, after
performing some file transfers I seem to lose network connectivity.

Please note that I do not need to stress the network card significantly
to lose network connectivity, transferring a file of roughly 20MB
appears to be sufficient.

I haven't tried a e1000 gigE nic at this time, let me know if other
options or workaround are available?


Thanks,
Jerome

----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-06-29 20:09

Message:
Logged In: YES 
user_id=1815
Originator: YES

Update: I was advised on irc to try e1000 network card emulation instead,
and with it I get very good results - 32 megabytes per second and more. So
the bug occurs only with rtl8139 emulation, perhaps because of bandwidth
limitation (packets aren't read quick enough from the TAP handle)?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 11:12
Message generated for change (Comment added) made by jibanes
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 02:04

Message:
Logged In: YES 
user_id=2174347
Originator: NO

It *seems* that a 'pcnet'-type doesn't shutdown the network emulation, can
you
repro? (I seem to have issues under Plan 9 with the e1000 nic).


Sincerely,
Jerome

----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 01:51

Message:
Logged In: YES 
user_id=2174347
Originator: NO

I've observed the same behavior with kvm-72 and 2.6.26.2 on intel,
the guest is a (very) up to date plan 9 operating system.

I have tried ne2k_pci AND rtl8139 emulation with tap, still, after
performing some file transfers I seem to lose network connectivity.

Please note that I do not need to stress the network card significantly
to lose network connectivity, transferring a file of roughly 20MB
appears to be sufficient.

I haven't tried a e1000 gigE nic at this time, let me know if other
options or workaround are available?


Thanks,
Jerome

----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-06-29 12:09

Message:
Logged In: YES 
user_id=1815
Originator: YES

Update: I was advised on irc to try e1000 network card emulation instead,
and with it I get very good results - 32 megabytes per second and more. So
the bug occurs only with rtl8139 emulation, perhaps because of bandwidth
limitation (packets aren't read quick enough from the TAP handle)?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 11:12
Message generated for change (Comment added) made by jibanes
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

Comment By: Jerome Ibanes (jibanes)
Date: 2008-08-09 01:51

Message:
Logged In: YES 
user_id=2174347
Originator: NO

I've observed the same behavior with kvm-72 and 2.6.26.2 on intel,
the guest is a (very) up to date plan 9 operating system.

I have tried ne2k_pci AND rtl8139 emulation with tap, still, after
performing some file transfers I seem to lose network connectivity.

Please note that I do not need to stress the network card significantly
to lose network connectivity, transferring a file of roughly 20MB
appears to be sufficient.

I haven't tried a e1000 gigE nic at this time, let me know if other
options or workaround are available?


Thanks,
Jerome

----------------------------------------------------------------------

Comment By: Andrew Zabolotny (andyz)
Date: 2008-06-29 12:09

Message:
Logged In: YES 
user_id=1815
Originator: YES

Update: I was advised on irc to try e1000 network card emulation instead,
and with it I get very good results - 32 megabytes per second and more. So
the bug occurs only with rtl8139 emulation, perhaps because of bandwidth
limitation (packets aren't read quick enough from the TAP handle)?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 19:12
Message generated for change (Comment added) made by andyz
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

>Comment By: Andrew Zabolotny (andyz)
Date: 2008-06-29 20:09

Message:
Logged In: YES 
user_id=1815
Originator: YES

Update: I was advised on irc to try e1000 network card emulation instead,
and with it I get very good results - 32 megabytes per second and more. So
the bug occurs only with rtl8139 emulation, perhaps because of bandwidth
limitation (packets aren't read quick enough from the TAP handle)?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

[ kvm-Bugs-2005957 ] TAP networking stalls on large file transfers

[SourceForge.net]

Bugs item #2005957, was opened at 2008-06-29 19:12
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: qemu
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Andrew Zabolotny (andyz)
Assigned to: Nobody/Anonymous (nobody)
Summary: TAP networking stalls on large file transfers

Initial Comment:
System: Fedora 9
KVM version: tried 65 and 70, both exhibit the same behaviour

When using TAP networking if I copy a large file from host to guest OS, I get extremely low performance (under 1 megabyte/s) and sometimes file transfer stalls. If I run tcpdump in guest OS (even with a bogus filter - e.g. tcpdump -nn -i eth0 host 1.2.3.4) the stalled connection suddenly come alive. If I run repeatedly tcpdump/press Ctrl+C and so on, the file transfer visually goes much faster.

I found a old thread about the same bug in QEMU: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290569

The patch proposed there was applied in QEMU 0.7.3, but when I looked into QEMU 0.9.1 sources - it is not there, so it seems it was removed for some reason. However, if I run the same virtual machine with QEMU, I get stable performance with about 8-9 megabytes per second flowing both ways (from host to guest and from guest to host).

I tried to apply the patch on KVM 70, but it does not make any difference :-(

The guest OS is Ubuntu 8.04 but I guess that does not matter.

I will be glad to cooperate to find/fix the bug but for now I'm out of ideas :-(


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=2005957&group_id=180599