于 2011年01月28日 23:33, Eric Blake 写道:
On 01/28/2011 05:11 AM, Osier Yang wrote:
于 2011年01月28日 20:02, Daniel P. Berrange 写道:
On Fri, Jan 28, 2011 at 07:53:44PM +0800, Osier Yang wrote:
As cgroup doesn't allow one writes negative into files like cpu.shares,
(e.g. echo -1> /cgroup/cpu/libvirt/qem
2011/1/28 Eric Blake :
> On 01/28/2011 02:16 PM, Matthias Bolte wrote:
>> ---
>> src/esx/esx_driver.c | 3 ++-
>> 1 files changed, 2 insertions(+), 1 deletions(-)
>>
>> diff --git a/src/esx/esx_driver.c b/src/esx/esx_driver.c
>> index 61c2371..c66df0e 100644
>> --- a/src/esx/esx_driver.c
>> +++
2011/1/28 Eric Blake :
> On 01/28/2011 02:39 PM, Matthias Bolte wrote:
>> Pure cosmetic change.
>
>> 24 files changed, 80 insertions(+), 96 deletions(-)
>
> ACK; hopefully it doesn't bite people trying to backport patches with
> too many spurious conflicts. At least it wasn't too many instances.
2011/1/28 Eric Blake :
> [replying to myself - never a good sign on Friday afternoon]
>
> On 01/28/2011 02:59 PM, Eric Blake wrote:
>>> +/* divide value by size, rounding up */
>>> +# define VIR_DIV_UP(value, size) (((value) + (size) - 1) / (size))
>>
>> Hmm - now that we are codifying it, can we m
On 01/28/2011 04:04 PM, Matthias Bolte wrote:
>> +++ b/.x-sc_prohibit_close
>> @@ -1,3 +1,9 @@
>> +# Non-C files:
>> ^docs/.*
>> +^ChangeLog*
>> ^HACKING$
>> +*.py$
>
> Shouldn't the dot be escaped here like in the other lines you added?
>
>> +# Wrapper implementation:
>> ^src/util/files.c$
>
* src/fdstream.c (virFDStreamOpenFile, virFDStreamCreateFile):
Use VIR_FORCE_CLOSE instead of close.
* tests/commandtest.c (mymain): Likewise.
* tools/virsh.c (editFile): Use virCommand instead of system.
* src/util/util.c (__virExec): Special case preservation of std
file descriptors to child.
---
2011/1/29 Eric Blake :
> On 01/28/2011 03:39 PM, Matthias Bolte wrote:
>>> + if (!editor)
>>> + editor = "vi"; /* could be cruel & default to ed(1) here */
>>
>> When VISUAL and EDITOR isn't set we fallback to vi here, but ...
>>
>>> + cmd = virCommandNewArgList("sh", "-c", NULL);
On 01/28/2011 03:39 PM, Matthias Bolte wrote:
>> +if (!editor)
>> +editor = "vi"; /* could be cruel & default to ed(1) here */
>
> When VISUAL and EDITOR isn't set we fallback to vi here, but ...
>
>> +cmd = virCommandNewArgList("sh", "-c", NULL);
>> +virCommandAddArgF
2011/1/28 Eric Blake :
> commit f1fe9671e was supposed to make sure we use files.h
> macros to avoid double close, but it didn't work.
>
> Meanwhile, virCommand is vastly superior to system(), fork(),
> and popen() (also to virExec, but we haven't completed that
> conversion), so enforce that, too.
[replying to myself - never a good sign on Friday afternoon]
On 01/28/2011 02:59 PM, Eric Blake wrote:
>> +/* divide value by size, rounding up */
>> +# define VIR_DIV_UP(value, size) (((value) + (size) - 1) / (size))
>
> Hmm - now that we are codifying it, can we make it less prone to
> overflow
2011/1/28 Eric Blake :
> * src/fdstream.c (virFDStreamOpenFile, virFDStreamCreateFile):
> Use VIR_FORCE_CLOSE instead of close.
> * tests/commandtest.c (mymain): Likewise.
> * tools/virsh.c (editFile): Use virCommand instead of system.
> ---
> src/fdstream.c | 6 ++--
> tests/commandtest.c
On 01/28/2011 03:26 PM, Matthias Bolte wrote:
> 2011/1/28 Eric Blake :
>> Fixes test failure that was overlooked after commit 1e1f7a8950.
>>
>> * daemon/Makefile.am (check-local): Let 'make check' fail on error.
>> * daemon/test_libvirtd.aug: Move qemu-specific option...
>> * src/qemu/test_libvirtd
2011/1/28 Eric Blake :
> Fixes test failure that was overlooked after commit 1e1f7a8950.
>
> * daemon/Makefile.am (check-local): Let 'make check' fail on error.
> * daemon/test_libvirtd.aug: Move qemu-specific option...
> * src/qemu/test_libvirtd_qemu.aug: ...into correct test.
> * src/qemu/libvirt
On 01/28/2011 11:21 AM, Michal Privoznik wrote:
> This will iterate over all NUMA nodes, showing
> free memory for each and sum at the end.
> Existing default behavior is not changed.
> +if (all_given && cell_given) {
> +vshError(ctl, "%s", _("--cellno and --all are mutually exclusive.
* src/fdstream.c (virFDStreamOpenFile, virFDStreamCreateFile):
Use VIR_FORCE_CLOSE instead of close.
* tests/commandtest.c (mymain): Likewise.
* tools/virsh.c (editFile): Use virCommand instead of system.
---
src/fdstream.c |6 ++--
tests/commandtest.c | 12 +++---
tools/virsh.c
commit f1fe9671e was supposed to make sure we use files.h
macros to avoid double close, but it didn't work.
Meanwhile, virCommand is vastly superior to system(), fork(),
and popen() (also to virExec, but we haven't completed that
conversion), so enforce that, too.
* cfg.mk (sc_prohibit_close): Fi
As reported here, 'make syntax-check' missed a case:
https://www.redhat.com/archives/libvir-list/2011-January/msg01165.html
This cleans up the remaining offenders, then improves the rules.
Eric Blake (2):
build: avoid close, system
maint: reject raw close, popen in 'make syntax-check'
.x-sc
On 01/28/2011 02:14 PM, Matthias Bolte wrote:
> Use it in all places where a memory or storage request size is converted
> to a larger granularity. This avoids requesting too small memory or storage
> sizes that could result from the truncation done by a simple division.
Much nicer than open-codin
On 01/28/2011 02:39 PM, Matthias Bolte wrote:
> Pure cosmetic change.
> 24 files changed, 80 insertions(+), 96 deletions(-)
ACK; hopefully it doesn't bite people trying to backport patches with
too many spurious conflicts. At least it wasn't too many instances.
--
Eric Blake ebl...@redhat.c
On 01/28/2011 02:16 PM, Matthias Bolte wrote:
> ---
> src/esx/esx_driver.c |3 ++-
> 1 files changed, 2 insertions(+), 1 deletions(-)
>
> diff --git a/src/esx/esx_driver.c b/src/esx/esx_driver.c
> index 61c2371..c66df0e 100644
> --- a/src/esx/esx_driver.c
> +++ b/src/esx/esx_driver.c
> @@ -20
Pure cosmetic change.
---
daemon/libvirtd.c |4 ++--
daemon/remote.c |4 ++--
src/conf/nwfilter_conf.c | 22 +++---
src/conf/nwfilter_conf.h |2 +-
src/conf/storage_conf.c
---
src/esx/esx_driver.c |3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/src/esx/esx_driver.c b/src/esx/esx_driver.c
index 61c2371..c66df0e 100644
--- a/src/esx/esx_driver.c
+++ b/src/esx/esx_driver.c
@@ -2059,8 +2059,9 @@ esxDomainSetMaxMemory(virDomainPtr domain, unsig
Fixes test failure that was overlooked after commit 1e1f7a8950.
* daemon/Makefile.am (check-local): Let 'make check' fail on error.
* daemon/test_libvirtd.aug: Move qemu-specific option...
* src/qemu/test_libvirtd_qemu.aug: ...into correct test.
* src/qemu/libvirtd_qemu.aug: Parse new option.
---
2011/1/27 Eric Blake :
> On 01/27/2011 02:35 PM, Matthias Bolte wrote:
>> Report an VIR_ERR_OPERATION_INVALID error in that case instead of letting
>> the SOAP call fail with an VIR_ERR_INTERNAL_ERROR error.
>> ---
>> src/esx/esx_driver.c | 21 ++---
>> 1 files changed, 18 insert
Hi there,
GSoC 2011 has been announced[1]. As we were pretty successful last year,
I think we should participate again. I've already created a wiki page:
http://wiki.qemu.org/Google_Summer_of_Code_2011
We should now populate it with projects and people willing to be mentors
should say so (or ju
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hello,
this is a draft of libvirt l10n/i18n translator's how-to. It is nowhere
definite and it actually is meant ...
It's practically summarized what we've talked about in last couple days.
~~~ DRAFT ~~~
# How-to libvirt l10n/i18n
We are glad you'v
Okay, I'm trying to understand the Python implentation (I'm not a python
expert but it seems to be quite readable once you've understand why there is
"self" everywhere :) ).
The thing I'm not able to understand is the call to the "poll" function as
this function doesn't exists in C# in any libr
This will iterate over all NUMA nodes, showing
free memory for each and sum at the end.
Existing default behavior is not changed.
---
tools/virsh.c | 75 +++-
1 files changed, 63 insertions(+), 12 deletions(-)
diff --git a/tools/virsh.c b/tool
On 01/24/2011 08:13 AM, Daniel P. Berrange wrote:
> A lock manager may operate in various modes. The direct mode of
> operation is to obtain locks based on the resources associated
> with devices in the XML. The indirect mode is where the app
> creating the domain provides explicit leases for each
On 01/24/2011 08:13 AM, Daniel P. Berrange wrote:
> When built as modules, the connection drivers live
> in $LIBDIR/libvirt/drivers. Now we add lock manager
> drivers, we need to distinguish. So move the existing
> modules to 'connection-driver'
>
> * src/Makefile.am: Move module install dir
> * s
On 01/24/2011 08:13 AM, Daniel P. Berrange wrote:
> To ensure child processes will log all error messages, reset
> the logging filter function when forking
>
> * src/util/util.c: Reset log filter in fork
> ---
> src/util/util.c |1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> dif
On 01/28/2011 09:42 AM, Michal Privoznik wrote:
> This will iterate over all NUMA nodes, showing
> free memory for each and sum at the end.
> Existing default behavior is not changed.
> ---
> tools/virsh.c | 73 +++-
> 1 files changed, 61 inser
On Fri, Jan 28, 2011 at 09:32:42AM -0700, Eric Blake wrote:
> On 01/24/2011 08:13 AM, Daniel P. Berrange wrote:
> > Some functionality run in virExec hooks may do I/O which
> > can trigger SIGPIPE. Renable SIGPIPE blocking around the
> > hook function
>
> > * src/util/util.c: Block SIGPIPE around
On 01/24/2011 08:13 AM, Daniel P. Berrange wrote:
> Some functionality run in virExec hooks may do I/O which
> can trigger SIGPIPE. Renable SIGPIPE blocking around the
> hook function
> * src/util/util.c: Block SIGPIPE around hooks
> -if (hook)
> +if (hook) {
> +/* virFork reset al
On 01/28/2011 08:19 AM, Daniel P. Berrange wrote:
> On Fri, Jan 14, 2011 at 03:58:10PM -0700, Eric Blake wrote:
>> qemu allows the user to choose what io storage api should be used,
>> either the default (threads) or native (linux aio) which in the latter
>> case can result in better performance.
>
This will iterate over all NUMA nodes, showing
free memory for each and sum at the end.
Existing default behavior is not changed.
---
tools/virsh.c | 73 +++-
1 files changed, 61 insertions(+), 12 deletions(-)
diff --git a/tools/virsh.c b/tool
On 01/26/2011 05:42 AM, Daniel P. Berrange wrote:
> On Sun, Jan 23, 2011 at 02:18:55PM +0800, Gui Jianfeng wrote:
>> Implement blkio.weight tuning API.
>>
>> Signed-off-by: Gui Jianfeng
>
> ACK
I saw that 1 and 2 had been ACK'd, and was about to push, but then I
looked further and saw that 3 had
On 01/28/2011 09:02 AM, Jim Meyering wrote:
> Eric Blake wrote:
>
>> On 01/28/2011 05:52 AM, Juerg Haefliger wrote:
>>> ---
>>> docs/page.xsl | 350
>>>
>>> 1 files changed, 175 insertions(+), 175 deletions(-)
>>
>> ACK, applied. po/ca.p
On 01/28/2011 08:59 AM, Eric Blake wrote:
>> So I pushed all the po from Fedora to git now, so the current state of
>> fedora and of git upstream should match (even if as told there will be
>> differences in the po files).
>
> Can you also ask the Fedora translation site to fix po/ca.po to quit
Eric Blake wrote:
> On 01/28/2011 05:52 AM, Juerg Haefliger wrote:
>> ---
>> docs/page.xsl | 350
>>
>> 1 files changed, 175 insertions(+), 175 deletions(-)
>
> ACK, applied. po/ca.po is the only other file with bad line endings,
> but t
On 01/26/2011 10:40 PM, Daniel Veillard wrote:
> Okay, in general the file will never match exactly, because I run
> "make update-po" in the po directory before commiting them to make sure
> thing like line numbers matches the current pot. I never push po back
> only the pot.
> So I pushed all
On 01/28/2011 05:52 AM, Juerg Haefliger wrote:
> ---
> docs/page.xsl | 350
> 1 files changed, 175 insertions(+), 175 deletions(-)
ACK, applied. po/ca.po is the only other file with bad line endings,
but that requires some coordination wi
Woohoo it will burn my brain :)
--
From: "Daniel P. Berrange"
Sent: Friday, January 28, 2011 4:24 PM
To:
Cc:
Subject: Re: [libvirt] Raising events for testing
On Fri, Jan 28, 2011 at 04:17:21PM +0100, arnaud.champ...@devatom.fr
wrote:
I suff
On Fri, Jan 28, 2011 at 08:33:20AM -0700, Eric Blake wrote:
> On 01/28/2011 05:11 AM, Osier Yang wrote:
> > 于 2011年01月28日 20:02, Daniel P. Berrange 写道:
> >> On Fri, Jan 28, 2011 at 07:53:44PM +0800, Osier Yang wrote:
> >>> As cgroup doesn't allow one writes negative into files like cpu.shares,
> >>
On Thu, Jan 27, 2011 at 09:50:38AM +0100, Daniel Lezcano wrote:
> On 01/27/2011 02:45 AM, Andrew Morton wrote:
> >On Thu, 27 Jan 2011 09:08:51 +0800 Li Zefan wrote:
> >
> >>Andrew Morton wrote:
> >>>On Tue, 25 Jan 2011 10:39:48 +0100
> >>>Daniel Lezcano wrote:
> >>>
> This patch removes the n
On 01/28/2011 05:11 AM, Osier Yang wrote:
> 于 2011年01月28日 20:02, Daniel P. Berrange 写道:
>> On Fri, Jan 28, 2011 at 07:53:44PM +0800, Osier Yang wrote:
>>> As cgroup doesn't allow one writes negative into files like cpu.shares,
>>> (e.g. echo -1> /cgroup/cpu/libvirt/qemu/rhel6/cpu.shares), user wil
On Fri, Jan 14, 2011 at 03:58:10PM -0700, Eric Blake wrote:
> qemu allows the user to choose what io storage api should be used,
> either the default (threads) or native (linux aio) which in the latter
> case can result in better performance.
>
> Based on a patch originally by Matthias Dahl.
>
>
On Fri, Jan 28, 2011 at 04:17:21PM +0100, arnaud.champ...@devatom.fr wrote:
> I suffer with the new callback API :)
>
> Does the new callback API need the EventRegisterImpl call ?
Yes, virEventRegisterImpl is required in order to be able to
receive events.
> Can anyone explain me the steps to i
I suffer with the new callback API :)
Does the new callback API need the EventRegisterImpl call ?
Can anyone explain me the steps to install these callbacks ? I have a lot of
difficulties to understand the polling process in the sample, does this
polling is mandatory ?
Help please
Arnaud
On Fri, Jan 14, 2011 at 03:58:09PM -0700, Eric Blake wrote:
> From: Matthias Dahl
>
> Signed-off-by: Eric Blake
> ---
> src/qemu/qemu_capabilities.c |2 ++
> src/qemu/qemu_capabilities.h |3 ++-
> tests/qemuhelptest.c | 12
> 3 files changed, 12 insertions(+), 5 d
On Fri, Jan 14, 2011 at 03:58:08PM -0700, Eric Blake wrote:
> From: Matthias Dahl
>
> Allows io={threads|native} as an optional attribute to .
>
> Signed-off-by: Eric Blake
> ---
> AUTHORS |1 +
> docs/formatdomain.html.in | 40 +---
>
Weird, only domain lifecycle events works :( I do something wrong...
--
From:
Sent: Friday, January 28, 2011 2:23 PM
To: "Daniel P. Berrange"
Cc:
Subject: Re: [libvirt] Raising events for testing
Many thanks Daniel ! I'll test that
--
---
docs/page.xsl | 350
1 files changed, 175 insertions(+), 175 deletions(-)
diff --git a/docs/page.xsl b/docs/page.xsl
index 4f1c92f..fc782a2 100644
--- a/docs/page.xsl
+++ b/docs/page.xsl
@@ -1,175 +1,175 @@
-
-http://www.w3.org/1999/XS
Many thanks Daniel ! I'll test that
--
From: "Daniel P. Berrange"
Sent: Friday, January 28, 2011 12:00 PM
To:
Cc:
Subject: Re: [libvirt] Raising events for testing
On Thu, Jan 27, 2011 at 11:50:05PM +0100, arnaud.champ...@devatom.fr
wrote:
?H
于 2011年01月28日 20:02, Daniel P. Berrange 写道:
On Fri, Jan 28, 2011 at 07:53:44PM +0800, Osier Yang wrote:
As cgroup doesn't allow one writes negative into files like cpu.shares,
(e.g. echo -1> /cgroup/cpu/libvirt/qemu/rhel6/cpu.shares), user will be
confused if libvirt accepts negative value and
On Fri, Jan 28, 2011 at 07:53:44PM +0800, Osier Yang wrote:
> As cgroup doesn't allow one writes negative into files like cpu.shares,
> (e.g. echo -1> /cgroup/cpu/libvirt/qemu/rhel6/cpu.shares), user will be
> confused if libvirt accepts negative value and converts it into unsigned
> int (or long i
As cgroup doesn't allow one writes negative into files like cpu.shares,
(e.g. echo -1> /cgroup/cpu/libvirt/qemu/rhel6/cpu.shares), user will be
confused if libvirt accepts negative value and converts it into unsigned
int (or long int, etc) silently.
* tools/virsh.c
---
tools/virsh.c |6 --
On Thu, Jan 27, 2011 at 11:50:05PM +0100, arnaud.champ...@devatom.fr wrote:
> ?Hi,
>
> I have binded the new API for domain events and callbacks in C#. It seems to
> work (LifeCycle events works for sure, I have tested them). Now I want to
> test other event types and I don't know how to raise t
?Hi,
I'm working on C# bindings. The goal is to obtain a more C# API for the API
users.
So I have create object for libvirt familly function (an object Connect,
Domain, Storage and so on).
All is almost done, but I need and advice about the way to handle object
properties. For example, for the
于 2011年01月27日 13:16, Gui Jianfeng 写道:
Hi
This patchset adds blkio cgroup support for qemu and lxc.
[PATCH 1/4] cgroup: Enable cgroup hierarchy for blkio cgroup
[PATCH 2/4 v2] cgroup: Implement blkio.weight tuning API.
[PATCH 3/4 v2] qemu: Implement blkio tunable XML configuration and parsing.
[
Matthias Bolte wrote:
> 2011/1/27 Gui Jianfeng :
>> Hi
>>
>> This patchset adds blkio cgroup support for qemu and lxc.
>>
>> [PATCH 1/4] cgroup: Enable cgroup hierarchy for blkio cgroup
>> [PATCH 2/4 v2] cgroup: Implement blkio.weight tuning API.
>> [PATCH 3/4 v2] qemu: Implement blkio tunable XML
2011/1/27 Gui Jianfeng :
> Hi
>
> This patchset adds blkio cgroup support for qemu and lxc.
>
> [PATCH 1/4] cgroup: Enable cgroup hierarchy for blkio cgroup
> [PATCH 2/4 v2] cgroup: Implement blkio.weight tuning API.
> [PATCH 3/4 v2] qemu: Implement blkio tunable XML configuration and parsing.
> [P
On Fri, 28 Jan 2011 15:19:11 +0800, Osier Yang wrote:
> 于 2011年01月28日 13:23, Nikunj A. Dadhania 写道:
> > From: Nikunj A. Dadhania
> >
> > Make cpu share persistent and add support for parsing them.
> >
> > docs/formatdomain.html.in: Document cputune element
> > src/conf/domain_conf.c,src/conf/domai
63 matches
Mail list logo