- Devel - libvirt List Archives

[Libvir] [PATCH] Cut out "type=ioemu" on Xen3.1
by Saori Fukuta 07 Jun '07

07 Jun '07

Hi, The definition of (type ioemu) is *not* required by Xen3.1. So I cut out it when xendConfigVersion is 4 or later. Signed-off-by: Saori Fukuta <fukuta.saori(a)jp.fujitsu.com> Thanks, Saori Fukuta. Index: xml.c =================================================================== RCS file: /data/cvs/libvirt/src/xml.c,v retrieving revision 1.73 diff -u -p -r1.73 xml.c --- xml.c 23 Apr 2007 07:41:23 -0000 1.73 +++ xml.c 31 May 2007 07:02:01 -0000 @@ -1047,6 +1047,7 @@ virDomainParseXMLDiskDesc(virConnectPtr * @conn: pointer to the hypervisor connection * @node: node containing the interface description * @buf: a buffer for the result S-Expr + * @xendConfigVersion: xend configuration file format * * Parse the one interface the XML description and add it to the S-Expr in buf * This is a temporary interface as the S-Expr interface @@ -1056,7 +1057,7 @@ virDomainParseXMLDiskDesc(virConnectPtr * Returns 0 in case of success, -1 in case of error. */ static int -virDomainParseXMLIfDesc(virConnectPtr conn ATTRIBUTE_UNUSED, xmlNodePtr node, virBufferPtr buf, int hvm) +virDomainParseXMLIfDesc(virConnectPtr conn ATTRIBUTE_UNUSED, xmlNodePtr node, virBufferPtr buf, int hvm, int xendConfigVersion) { xmlNodePtr cur; xmlChar *type = NULL; @@ -1129,7 +1130,8 @@ virDomainParseXMLIfDesc(virConnectPtr co virBufferVSprintf(buf, "(script '%s')", script); if (ip != NULL) virBufferVSprintf(buf, "(ip '%s')", ip); - if (hvm) + /* HVM type ioemu config for xen < 3.1 */ + if (hvm && xendConfigVersion < 4) virBufferAdd(buf, "(type ioemu)", 12); virBufferAdd(buf, ")", 1); @@ -1335,7 +1337,7 @@ virDomainParseXMLDesc(virConnectPtr conn if (nb_nodes > 0) { for (i = 0; i < nb_nodes; i++) { virBufferAdd(&buf, "(device ", 8); - res = virDomainParseXMLIfDesc(conn, nodes[i], &buf, hvm); + res = virDomainParseXMLIfDesc(conn, nodes[i], &buf, hvm, xendConfigVersion); if (res != 0) { free(nodes); goto error; @@ -1494,7 +1496,7 @@ virParseXMLDevice(virConnectPtr conn, ch goto error; } else if (xmlStrEqual(node->name, BAD_CAST "interface")) { - if (virDomainParseXMLIfDesc(conn, node, &buf, hvm) != 0) + if (virDomainParseXMLIfDesc(conn, node, &buf, hvm, xendConfigVersion) != 0) goto error; } cleanup:

3 6

[Libvir] Add scheduler patch?
by Atsushi SAKAI 06 Jun '07

06 Jun '07

Hi, Is there any plan to apply scheduler controller patch? https://www.redhat.com/archives/libvir-list/2007-May/msg00291.html https://www.redhat.com/archives/libvir-list/2007-May/msg00295.html Thanks Atsushi SAKAI

2 12

[Libvir] [PATCH] make xen compile -Werror and without PROXY turned on (the autobuild config)
by Hugh Brock 05 Jun '07

05 Jun '07

$SUBJECT says it all... --Hugh -- Red Hat Virtualization Group http://redhat.com/virtualization Hugh Brock | virt-manager http://virt-manager.org hbrock(a)redhat.com | virtualization library http://libvirt.org

2 1

Re: [Libvir] [PATCH] add scheduler API(take 3?)
by Richard W.M. Jones 04 Jun '07

04 Jun '07

+char * +virDomainGetSchedulerType(virDomainPtr domain, int *nparams) +{ + virConnectPtr conn; + char *schedtype; + + if (domain == NULL) { + TODO + return NULL; + } Is this case an error, or were you thinking of adding more semantics later on here? (and the same comment for virDomainGet/SetSchedulerParameters) +static int +cmdSchedinfo(vshControl * ctl, vshCmd * cmd) +{ [...] + char *str_weight = strdup("weight"); + char *str_cap = strdup("cap"); + + nparams = malloc(sizeof(int)); + *nparams = 0; + + if (!vshConnectionUsability(ctl, ctl->conn, TRUE)) + return FALSE; There's still a problem memory leak here. + /* Currently supports Xen Credit only */ + weight = vshCommandOptInt(cmd, "weight", &weightfound); + if(weightfound){ inputparam++; } + + cap = vshCommandOptInt(cmd, "cap", &capfound); + if(capfound){ inputparam++; } and can this be made so it isn't Xen-specific? + if ((domain == NULL) || (domain->conn == NULL)) + return -1; + + priv = (xenUnifiedPrivatePtr) domain->conn->privateData; + if (priv->handle < 0 || domain->id < 0) + return -1; At the lowest level, these functions should return error messages back to the upper layers. Otherwise users have nothing to diagnose errors with. + /* + * Support only dom_interface_version >=5 + * (Xen3.1.0 or later) + */ + if (dom_interface_version < 5) + return -1; Is this because earlier hypervisors didn't support this, or is it just not implemented? Rich.

3 7

[Libvir] [PATCH] Fix strange error message in virsh attach-device
by Masayuki Sunou 04 Jun '07

04 Jun '07

Hi Now, virsh attach-device displays the following messages when "device" attribute of "disk" element is "floppy" in XML. # virsh attach-device HVM_RH4_file floppy.xml libvir: Xen Daemon error : POST operation failed: (xend.err 'Invalid Configuration: XendConfig: @\x14\xce not a valid device type') error: Failed to attach device from floppy.xml This patch fixes this strange error message displaying "@\x14\xce". Description of this patch VirDomainParseXMLDiskDesc() returns 0 without putting data in virBuffer, when "floppy" is set. Therefore, virParseXMLDevice() returns no initialized data, and xenDaemonAttachDevice() passes Xen this data. Because virDomainParseXMLDesc() works wrong when this patch changes return value of VirDomainParseXMLDiskDesc(),this patch fixes virParseXMLDevice() as follows. - Initialize data - Return error if data does not change. As a result, xenDaemonAttachDevice() doesn't pass strange data to Xen, and Xen doesn't return strange error message. Signed-off-by: Masayuki Sunou <fj1826dm(a)aa.jp.fujitsu.com> Thanks, Masayuki Sunou. ---------------------------------------------------------------------- Index: src/xml.c =================================================================== RCS file: /data/cvs/libvirt/src/xml.c,v retrieving revision 1.73 diff -u -p -r1.73 xml.c --- src/xml.c 23 Apr 2007 07:41:23 -0000 1.73 +++ src/xml.c 25 May 2007 09:46:56 -0000 @@ -1481,6 +1481,7 @@ virParseXMLDevice(virConnectPtr conn, ch return (NULL); buf.size = 1000; buf.use = 0; + memset(buf.content, 0, (size_t)buf.size); xml = xmlReadDoc((const xmlChar *) xmldesc, "domain.xml", NULL, XML_PARSE_NOENT | XML_PARSE_NONET | XML_PARSE_NOERROR | XML_PARSE_NOWARNING); @@ -1492,6 +1493,9 @@ virParseXMLDevice(virConnectPtr conn, ch if (xmlStrEqual(node->name, BAD_CAST "disk")) { if (virDomainParseXMLDiskDesc(conn, node, &buf, hvm, xendConfigVersion) != 0) goto error; + /* SXP is not created when device is "floppy". */ + else if (!strlen(buf.content)) + goto error; } else if (xmlStrEqual(node->name, BAD_CAST "interface")) { if (virDomainParseXMLIfDesc(conn, node, &buf, hvm) != 0) ----------------------------------------------------------------------

2 5

[Libvir] Use virsh to monitor qemud
by Fabian Deutsch 03 Jun '07

03 Jun '07

Hey, I jsut installed Fedora 7 and tried to install some guests and it worked like a charm. I use qemu for full virtulization. A couple of minutes later i stumbled across libvirts virtsh. But now I wonder how I cna use virtsh to connect to qemud - all I get is something like: "Can't connect to hyperisor." - fabiand

3 4

[Libvir] SOLVED Re: cannot use vncviewer and VMM graphics console to access HVM guest on RHES 5
by Glen Deem 03 Jun '07

03 Jun '07

Thanks to Igor Chubin (spasibo) and Richard Jones, the qemu vnc server was listening but I did not connect to its proper port (why libvirt doesn't use <IP>:<domU id> kind of vnc connection and restricts the server to local host only by default?). Regards, Glen On 5/31/07, Glen Deem <xen.inbox@XXXXXXXXX> wrote: > Hi all, > > I've created an HVM guest under RHES 5 (Xen 3.0.3) using "xm create" > but I cannot access it with vncviewer. > > The Virtual Machine Manager also informs about graphics console unavailability. > > What am I doing wrong, please? > > I do > > vncviewer 123.456.789.123:4, > where 4 is the guest id. > > This is the xm config file is: > > name = "hvmGuest" > builder = "hvm" > memory = "500" > disk = [ 'file:/var/lib/red-hat.img,hda,w', ] > vif = [ 'type=ioemu, mac=00:16:3e:34:3b:ba, bridge=xenbr0', ] > device_model = "/usr/lib/xen/bin/qemu-dm" > kernel = "/usr/lib/xen/boot/hvmloader" > vnc=1 > vncunused=1 > apic=1 > acpi=1 > pae=1 > vcpus=1 > serial = "pty" > on_reboot = 'restart' > on_crash = 'restart' > > Thanks a lot in advance. > > Glen from Ottawa, Canada >

4 4

[Libvir] Re: [Xen-users] SOLVED Re: cannot use vncviewer and VMM graphics console to access HVM guest on RHES 5
by Glen Deem 01 Jun '07

01 Jun '07

Hi Dan, I'm with you about the VNC authentication. But I still don't understand the logic behind assigning HVM displays. Instead of clear "domU ID is the display number" (vncunused 0), the Virtual Machine Manager assigns vncunused to 1 and then calculates the 5900+ port, I suppose depending on what port is not occupied (BTW where exactly is this code in the source, please?). Why is that? And how am I suppose to find out what that port number is if I'd want to connect to a VMM-created vm manually with my vnc client? Thanks, Glen Daniel P. Berrange wrote: > On Fri, Jun 01, 2007 at 11:03:36AM -0400, Glen Deem wrote: >> Thanks to Igor Chubin (spasibo) and Richard Jones, the qemu vnc server >> was listening but I did not connect to its proper port (why libvirt >> doesn't use <IP>:<domU id> kind of vnc connection and restricts the >> server to local host only by default?). > > VNC authentication is an utter joke. It can be trivially brute forced > so exposing it on a public IP address is not a good idea, hence the > default is 127.0.0.1, though even that's not ideal because it is still > exposed to local users. Ultimately VNC needs to have SSL/TLS support > integrated into it to allow secure access over public network, which > is something I'm working on for QEMU... > > Dan.

2 1

[Libvir] cannot use vncviewer and VMM graphics console to access HVM guest on RHES 5
by Glen Deem 01 Jun '07

01 Jun '07

Hi all, I've created an HVM guest under RHES 5 (Xen 3.0.3) using "xm create" but I cannot access it with vncviewer. The Virtual Machine Manager also informs about graphics console unavailability. What am I doing wrong, please? I do > vncviewer 123.456.789.123:4, where 4 is the guest id. This is the xm config file is: name = "hvmGuest" builder = "hvm" memory = "500" disk = [ 'file:/var/lib/red-hat.img,hda,w', ] vif = [ 'type=ioemu, mac=00:16:3e:34:3b:ba, bridge=xenbr0', ] device_model = "/usr/lib/xen/bin/qemu-dm" kernel = "/usr/lib/xen/boot/hvmloader" vnc=1 vncunused=1 apic=1 acpi=1 pae=1 vcpus=1 serial = "pty" on_reboot = 'restart' on_crash = 'restart' Thanks a lot in advance. Glen from Ottawa, Canada

3 2

Re: [Libvir] Debian package structure
by Marco Sinhoreli 31 May '07

31 May '07

On 5/30/07, Richard W.M. Jones <rjones(a)redhat.com> wrote: > > Marco Sinhoreli wrote: > > hello all, > > > > I have been worked in libvirt, virt-manager and virtinst packages to > > Debian for a weeks. Packages are now stabilized. Following below the > > /debian directories structure for analise. > > * libvirt: > > svn co http://svn.mussicorp.net/projetos/src/xen/libvirt/trunk/ > > In debian/rules you shouldn't need this any longer (with latest CVS): > > export CFLAGS += -fno-stack-protector I know it is fixed for libvirt, but it is nedded to fix virt-manager too. That's why I keep it in this package. > * Virt-manager: > > svn co > > > http://svn.mussicorp.net/projetos/src/xen/virt-manager/virt-manager/trunk/ > > Shouldn't virt-manager Depends: on python-libvirt and virtinst? My fault. I'll fix it. It should probably Build-Depends: on (eg) cdbs too, since firstly you're > using cdbs in debian/rules, and secondly virt-manager isn't pure Python > - it contains a C-based widget. okay. I'll pay attention on that. Not clear what the postrm was for. Is there configuration kept in > /usr/share/virt-manager? My fault. I'll checked it. Thanks

1 0

[Libvir] Debian package structure
by Marco Sinhoreli 30 May '07

30 May '07

hello all, I have been worked in libvirt, virt-manager and virtinst packages to Debian for a weeks. Packages are now stabilized. Following below the /debian directories structure for analise. * libvirt: svn co http://svn.mussicorp.net/projetos/src/xen/libvirt/trunk/ * Virt-manager: svn co http://svn.mussicorp.net/projetos/src/xen/virt-manager/virt-manager/trunk/ * Virtinst: svn co http://svn.mussicorp.net/projetos/src/xen/virt-manager/virtinst/trunk/ Best regards, -- Marco Sinhoreli

2 1

Re: [Libvir] Error while run make: undefined reference to `xenHypervisorMakeCapabilitiesXML'
by Marco Sinhoreli 30 May '07

30 May '07

Thanks, I had a problem with my libxen package. The /usr/lib/libxenstore.so wasn't linked to /usr/lib/libxenstore.so.3.0.0. Regards, > > On 5/29/07, Daniel Veillard <veillard(a)redhat.com> wrote: > > > > On Mon, May 28, 2007 at 11:26:18AM -0400, Daniel Veillard wrote: > > > On Mon, May 28, 2007 at 12:15:31PM -0300, Marco Sinhoreli wrote: > > > > Hello list, > > > > > > > > I'm with this error while I am compiling the libvirt: > > > > > > > > xencapstest.o: In function `testCompareFiles': > > > > /usr/src/libvirt/tests/xencapstest.c:33: undefined reference to > > > > `xenHypervisorMakeCapabilitiesXML' > > > > collect2: ld returned 1 exit status > > > > > > > > Any idea? > > > > > > What version ? I guess it comes when compiling on a machine where > > > Xen developments environment is not present, this ought to be fixed > > > if not already in CVS, > > > > actually I can reproduce that to on CVs head, will fix ! > > > > Daniel > > > > -- > > Red Hat Virtualization group http://redhat.com/virtualization/ > > Daniel Veillard | virtualization library http://libvirt.org/ > > veillard(a)redhat.com | libxml GNOME XML XSLT toolkit > > http://xmlsoft.org/ > > http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/ > > > > > > -- > Marco Sinhoreli -- Marco Sinhoreli

1 0

Re: [Libvir] Finally - fix the build on Debian
by Marco Sinhoreli 30 May '07

30 May '07

Hello, I compiled the libvirt from last cvs source without to export CFLAGS -fstack-protector option without problem. After it, I compiled the virt-manger and it isn't working. Running virt-manager retur this error: virt-manager Traceback (most recent call last): File "/usr/share/virt-manager/virt-manager.py", line 104, in ? from virtManager.config import vmmConfig File "/usr/share/virt-manager/virtManager/config.py", line 24, in ? import libvirt File "/usr/lib/python2.4/site-packages/libvirt.py", line 7, in ? import libvirtmod ImportError: /usr/lib/libvirt.so.0: undefined symbol: __stack_chk_guard Also is necessáry correction in virt-manager code? regards, On 5/29/07, Richard W.M. Jones <rjones(a)redhat.com> wrote: > > > Daniel Veillard wrote: > > > On Tue, May 29, 2007 at 03:21:48PM +0100, Richard W.M. Jones wrote: > > >> The attached patch fixes three separate build problems on Debian: > > >> > > >> (1) The libtool autoconf macro is wrong in that it requires a C++ > > >> compiler and /lib/cpp, neither of which seems to be really needed. > > >> > > >> (2) If you pass -fstack-protector while compiling, then you also need > > to > > >> pass it while linking, so that gcc knows to link -lssp. > > >> > > >> (3) Long-standing problem that xencapstest.c won't compile > > --without-xen. > > >> > > >> Well, OK so I didn't actually fix (1), but I did add a note to the > > TODO > > >> file pointing out the problem. So you'll still need to install the > > >> Debian "g++" package before you can compile libvirt. > > > > > > None of this is actually Debian specific, it actually affects various > > setups. > > > Patch looks fine to me, > > > > Committed. > > > > Rich. > > > > -- > > Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/<http://et.redhat.com/%7Erjones/> > > Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod > > Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in > > England and Wales under Company Registration No. 03798903 > > > > -- > > Libvir-list mailing list > > Libvir-list(a)redhat.com > > https://www.redhat.com/mailman/listinfo/libvir-list > > > > > > > > > -- > Marco Sinhoreli -- Marco Sinhoreli

1 0

[Libvir] Finally - fix the build on Debian
by Richard W.M. Jones 29 May '07

29 May '07

The attached patch fixes three separate build problems on Debian: (1) The libtool autoconf macro is wrong in that it requires a C++ compiler and /lib/cpp, neither of which seems to be really needed. (2) If you pass -fstack-protector while compiling, then you also need to pass it while linking, so that gcc knows to link -lssp. (3) Long-standing problem that xencapstest.c won't compile --without-xen. Well, OK so I didn't actually fix (1), but I did add a note to the TODO file pointing out the problem. So you'll still need to install the Debian "g++" package before you can compile libvirt. Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903

2 2

[Libvir] Error while run make: undefined reference to `xenHypervisorMakeCapabilitiesXML'
by Marco Sinhoreli 29 May '07

29 May '07

Hello list, I'm with this error while I am compiling the libvirt: xencapstest.o: In function `testCompareFiles': /usr/src/libvirt/tests/xencapstest.c:33: undefined reference to `xenHypervisorMakeCapabilitiesXML' collect2: ld returned 1 exit status Any idea? Regards. -- Marco Sinhoreli

2 2

[Libvir] [PATCH] Fix some memory leak
by Masayuki Sunou 29 May '07

29 May '07

Hi This patch fixes some memory leak in libvirt. 1. The point that checks command-option in virsh.c 2. The point that checks version of Xen in xen_internal.c 3. The point that checks OS type of VM in xend_internal.c And these memory leaks can be checked by valgrind. Signed-off-by: Masayuki Sunou <fj1826dm(a)aa.jp.fujitsu.com> Thanks, Masayuki Sunou. ------------------------------------------------------------------------------- Index: src/virsh.c =================================================================== RCS file: /data/cvs/libvirt/src/virsh.c,v retrieving revision 1.76 diff -u -p -r1.76 virsh.c --- src/virsh.c 23 May 2007 15:09:19 -0000 1.76 +++ src/virsh.c 29 May 2007 07:43:48 -0000 @@ -3153,8 +3153,10 @@ vshCommandParse(vshControl * ctl, char * c->def = cmd; c->next = NULL; - if (!vshCommandCheckOpts(ctl, c)) + if (!vshCommandCheckOpts(ctl, c)) { + if(c) free(c); goto syntaxError; + } if (!ctl->cmd) ctl->cmd = c; Index: src/xen_internal.c =================================================================== RCS file: /data/cvs/libvirt/src/xen_internal.c,v retrieving revision 1.72 diff -u -p -r1.72 xen_internal.c --- src/xen_internal.c 30 Apr 2007 17:30:11 -0000 1.72 +++ src/xen_internal.c 29 May 2007 07:43:48 -0000 @@ -1497,11 +1497,15 @@ xenHypervisorInit(void) virXenError(VIR_ERR_XEN_CALL, " ioctl ", IOCTL_PRIVCMD_HYPERCALL); close(fd); in_init = 0; + if (ipt) + free(ipt); return(-1); done: close(fd); in_init = 0; + if (ipt) + free(ipt); return(0); } Index: src/xend_internal.c =================================================================== RCS file: /data/cvs/libvirt/src/xend_internal.c,v retrieving revision 1.114 diff -u -p -r1.114 xend_internal.c --- src/xend_internal.c 11 May 2007 14:21:28 -0000 1.114 +++ src/xend_internal.c 29 May 2007 07:43:49 -0000 @@ -3072,7 +3072,7 @@ xenDaemonCreateLinux(virConnectPtr conn, static int xenDaemonAttachDevice(virDomainPtr domain, char *xml) { - char *sexpr, *conf; + char *sexpr, *conf, *str; int hvm = 0, ret; xenUnifiedPrivatePtr priv; @@ -3084,8 +3084,11 @@ xenDaemonAttachDevice(virDomainPtr domai priv = (xenUnifiedPrivatePtr) domain->conn->privateData; - if (strcmp(virDomainGetOSType(domain), "linux")) + str = virDomainGetOSType(domain); + if (strcmp(str, "linux")) hvm = 1; + if (str) + free(str); sexpr = virParseXMLDevice(domain->conn, xml, hvm, priv->xendConfigVersion); if (sexpr == NULL) return (-1); -------------------------------------------------------------------------------

2 1

[Libvir] [PATCH] add scheduler API(take 3?)
by Atsushi SAKAI 29 May '07

29 May '07

Hi, This patch is retry from one-month ago. This patch is for fixes following suggestions. https://www.redhat.com/archives/libvir-list/2007-April/msg00090.html And I add Check Xen3.1.0 only. (As you know domctl on Xen3.0.4 does not work fine.) How to use is described in following mail. https://www.redhat.com/archives/libvir-list/2007-April/msg00050.html Signed-off-by: Atsushi SAKAI <sakaia(a)jp.fujitsu.com> Thanks Atsushi SAKAI

5 7

[Libvir] [PATCH] Add Python binding for virGetVersion
by Richard W.M. Jones 29 May '07

29 May '07

This patch adds a Python binding for the virGetVersion call (called libvirt.getVersion). $ python Python 2.4.4 (#1, Oct 23 2006, 13:58:18) [GCC 4.1.1 20061011 (Red Hat 4.1.1-30)] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import libvirt; >>> libvirt.getVersion (None); (2002, 3000001) >>> libvirt.getVersion ("QEMU"); (2002, 2002) It also gives you a way to detect if a particular driver is compiled in, which I guess could be quite useful: >>> libvirt.getVersion ("not_here"); libvir: error : no support for hypervisor Traceback (most recent call last): File "<stdin>", line 1, in ? File "/home/rjones/local/lib/python2.4/site-packages/libvirt.py", line 95, in getVersion if ret is None: raise libvirtError ("virGetVersion() failed") libvirt.libvirtError: virGetVersion() failed no support for hypervisor Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903

3 7

[Libvir] PATCH: Create a logfile for each QEMU vm
by Daniel P. Berrange 29 May '07

29 May '07

Most of the error reporting the QEMU daemon is focused on sending errors wrt to RPC ops to the client. There is no provision currently for getting info about the QEMU vm itself when problems occur. The most common problem is some mis-configurable in the XML causes qemu to fail to start up. This results in the completely accurate & completely useless (mea-culpa) error message libvirt.libvirtError: virDomainCreateLinux() failed internal error End-of-file while reading PTY startup output' What this means is that qemu crashed/exited at startup. Now this patch does not (yet) directly address that stupid error message, but what it does do is provide a way to diagnose the real problem. For every VM we start it will create a logfile /etc/libvirt/qemu/logs/[vmname].log Or $HOME/.libvirt/qemu/logs/[vmname].log And all output from QEMU's stderr/stdout will get written to this file. So # virsh --connect qemu:///system start wizz libvir: QEMU error : internal error End-of-file while reading PTY startup output error: Failed to start domain wizz # cat /etc/libvirt/qemu/logs/wizz.log qemu: could not open hard disk image '/home/berrange/src/xen/virtinst--devel/demo' Next up I'm going to work on a) Improving the error message b) Adding validation that disk images actually exist before launching QEMU since that's the most common error we're seeing Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

6 9

[Libvir] AC_PROG_CPP
by Richard W.M. Jones 29 May '07

29 May '07

We have AC_PROG_CPP, but apparently no C++ code? In any case unless a C++ compiler is installed, I get: checking whether we are using the GNU C++ compiler... no checking whether g++ accepts -g... no checking dependency style of g++... none checking how to run the C++ preprocessor... /lib/cpp configure: error: C++ preprocessor "/lib/cpp" fails sanity check See `config.log' for more details. Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903

3 4

[Libvir] Question on acquiring cpuTime in struct _virDomainInfo
by Jan Michael 29 May '07

29 May '07

Hi everyone, using libvirt I'm trying to calculate cpu utilization of a node in percent. But sometimes values beyond 100.0% are being calculated. This is because a domain spend more time on a cpu than time is elapsed in the meantime. A short explanation of the way how cpu utilization is computed in my case: 1. - open two connections with conn_cur/conn_old = virConnectOpenReadOnly(NULL); 2. - get current time gettimeofday(&time_old, NULL); - get domain by id with dom_old = virDomainLookupByID(conn_old, id) - get domain information virDomainGetInfo(dom_old, &info_old); 3. - sleep a second 4. - doing same stuff like in 2. but with _cur 5. - compute cpu utilization by dividing used cputime by elapsed time and multiply with 100 Am I right if I suppose that cpuTime for _virDomainInfo structure will be directly acquired from the hypervisor in virDomainGetInfo (dom_old, &info_old) or is it already present with getting the domain itself? Is there any better solution of doing this, which is more precise? And another general question: The monitoring utility of xen, called xentop, provides also statistics about networking and vbds. Are there any plans to provide this values by libvirt in the future? Cheers, Jan

3 5

[Libvir] Please check my autoconf libvirt.m4
by Richard W.M. Jones 24 May '07

24 May '07

Attached is an autoconf snippet which allows you to check that the libvirt library is installed (not necessarily development headers though) and check the version number. If any autoconf experts (hello, Mark) would like to check it makes sense, that'd be great. Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903

4 7

[Libvir] Remote patch, 2007-05-22
by Richard W.M. Jones 22 May '07

22 May '07

This is the latest version of the remote patch. http://annexia.org/tmp/libvirt-remote-20070522.patch I'd like to propose this for merging. The only differences are those described here: https://www.redhat.com/archives/libvir-list/2007-May/msg00156.html, and a bit more testing. Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903

1 0

Re: [Libvir] [RFC]OpenVZ XML def
by Richard W.M. Jones 22 May '07

22 May '07

Shuveb Hussain wrote: >> There is the same problem with the Xen libraries. Dan had to rewrite a >> lot of code concerning access to the hypervisor. > > No matter how different I think, the end result looks just like the > OpenVZ library code :-P > So, I'm trying to write code that looks different, but does the same > thing. The OpenVZ utilities are however, scripting friendly. Their > text output is consistent. How about writing a daemon for OpenVZ that > "talks" to the OpenVZ utilities, rather than trying to re-write the > utilities? Would this be a bad idea? It sounds like a very good idea. There can also be no question of infringement if done this way. > The actual libvirt can then > connect to this daemon to get the real stuff done. This is what I did > in my EasyVZ GUI tool for OpenVZ. Why is the daemon needed though? The libvirt calls can just launch the OpenVZ command line tools as necessary. We got into daemons with QEMU, and it's nothing but trouble. Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903

1 0

[Libvir] Re: [Fedora-xen] BUG at arch/x86_64/mm/../../i386/mm/hypervisor.c:185!
by Richard W.M. Jones 22 May '07

22 May '07

Naoki wrote: > On Mon, 2007-05-21 at 18:53 +0100, Richard W.M. Jones wrote: >> I've been stressing Xen all week, and found plenty of bugs but not this >> one. Can I ask a few questions about your set-up? > > Fire away.. > >> - Is this dmesg in the dom0 or the domU (guest)? > > dom0 > >> - You seem to have an x86-64 SMP kernel. Can you provide details of >> the machine (how many cores, how much RAM, etc.) > > Single chip, dual core, intel Core 2 Duo. > > Family: Pentium 4 > Manufacturer: Intel > ID: 47 0F 00 00 FF FB EB BF > Signature: Type 0, Family 15, Model 4, Stepping 7 > > >> - How many guests are you running, and of what sort. eg. What is the >> guest OS? Is it paravirt or full-virt? > > I had one inactive guest running. > dom0 is running x86_64 kernel 2.6.20-2925.8.fc7xen. > And I'm on rawhide fully updated. That's the same set-up I'm using. Do you get this problem often? Is it reproducible? Is there a particular thing that you are doing when this happens? Rich. -- Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/ Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 03798903

2 2