October 2013 - Devel - libvirt List Archives

[libvirt] [PATCH] Improve debugging of QEMU start/stop
by Daniel P. Berrange 31 Oct '13

31 Oct '13

From: "Daniel P. Berrange" <berrange(a)redhat.com> Include reference of the VM object pointer and name in debug logs for QEMU start/stop functions. Also make sure we log the PID that we started, since it isn't available elsewhere in the logs. Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/qemu/qemu_process.c | 33 ++++++++++++--------------------- 1 file changed, 12 insertions(+), 21 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index bdffdf8..dd7566b 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -3499,6 +3499,9 @@ int qemuProcessStart(virConnectPtr conn, virQEMUDriverConfigPtr cfg; virCapsPtr caps = NULL; + VIR_DEBUG("vm=%p name=%s id=%d pid=%d", + vm, vm->def->name, vm->def->id, vm->pid); + /* Okay, these are just internal flags, * but doesn't hurt to check */ virCheckFlags(VIR_QEMU_PROCESS_START_COLD | @@ -3845,23 +3848,11 @@ int qemuProcessStart(virConnectPtr conn, _("Domain %s didn't show up"), vm->def->name); ret = -1; } -#if 0 - } else if (ret == -2) { - /* - * XXX this is bogus. It isn't safe to set vm->pid = child - * because the child no longer exists. - */ - - /* The virCommand process that launches the daemon failed. Pending on - * when it failed (we can't determine for sure), there may be - * extra info in the domain log (if the hook failed for example). - * - * Pretend like things succeeded, and let 'WaitForMonitor' report - * the log contents for us. - */ - vm->pid = child; - ret = 0; -#endif + VIR_DEBUG("QEMU vm=%p name=%s running with pid=%llu", + vm, vm->def->name, (unsigned long long)vm->pid); + } else { + VIR_DEBUG("QEMU vm=%p name=%s failed to spawn", + vm, vm->def->name); } VIR_DEBUG("Writing early domain status to disk"); @@ -4085,8 +4076,8 @@ qemuProcessKill(virDomainObjPtr vm, unsigned int flags) { int ret; - VIR_DEBUG("vm=%s pid=%d flags=%x", - vm->def->name, vm->pid, flags); + VIR_DEBUG("vm=%p name=%s pid=%d flags=%x", + vm, vm->def->name, vm->pid, flags); if (!(flags & VIR_QEMU_PROCESS_KILL_NOCHECK)) { if (!virDomainObjIsActive(vm)) { @@ -4126,8 +4117,8 @@ void qemuProcessStop(virQEMUDriverPtr driver, char ebuf[1024]; virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver); - VIR_DEBUG("Shutting down VM '%s' pid=%d flags=%x", - vm->def->name, vm->pid, flags); + VIR_DEBUG("Shutting down vm=%p name=%s id=%d pid=%d flags=%x", + vm, vm->def->name, vm->def->id, vm->pid, flags); if (!virDomainObjIsActive(vm)) { VIR_DEBUG("VM '%s' not active", vm->def->name); -- 1.8.3.1

2 1

[libvirt] [PATCH] storage: use correct type for array count
by Eric Blake 31 Oct '13

31 Oct '13

Using size_t counts will let us use VIR_APPEND_ELEMENT and friends. * src/conf/storage_conf.h (_virStoragePoolObjList) (_virStorageVolDefList): Track list sizes with size_t. * src/storage/storage_backend_rbd.c (virStorageBackendRBDRefreshPool): Fix type fallout. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- src/conf/storage_conf.h | 4 ++-- src/storage/storage_backend_rbd.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h index 251e897..d13cb99 100644 --- a/src/conf/storage_conf.h +++ b/src/conf/storage_conf.h @@ -112,7 +112,7 @@ struct _virStorageVolDef { typedef struct _virStorageVolDefList virStorageVolDefList; typedef virStorageVolDefList *virStorageVolDefListPtr; struct _virStorageVolDefList { - unsigned int count; + size_t count; virStorageVolDefPtr *objs; }; @@ -341,7 +341,7 @@ struct _virStoragePoolObj { typedef struct _virStoragePoolObjList virStoragePoolObjList; typedef virStoragePoolObjList *virStoragePoolObjListPtr; struct _virStoragePoolObjList { - unsigned int count; + size_t count; virStoragePoolObjPtr *objs; }; diff --git a/src/storage/storage_backend_rbd.c b/src/storage/storage_backend_rbd.c index 4e18bf9..b7a41c2 100644 --- a/src/storage/storage_backend_rbd.c +++ b/src/storage/storage_backend_rbd.c @@ -381,7 +381,7 @@ static int virStorageBackendRBDRefreshPool(virConnectPtr conn, pool->volumes.objs[pool->volumes.count++] = vol; } - VIR_DEBUG("Found %d images in RBD pool %s", + VIR_DEBUG("Found %zu images in RBD pool %s", pool->volumes.count, pool->def->source.name); ret = 0; -- 1.8.3.1

2 2

[libvirt] QEMU 1.6 and drive discard parameter
by whitearchey 31 Oct '13

31 Oct '13

In QEMU 1.6 parameters of 'drive' option were removed: QemuOptsList qemu_drive_opts = { .name = "drive", .head = QTAILQ_HEAD_INITIALIZER(qemu_drive_opts.head), .desc = { /* * no elements => accept any params * validation will happen later */ { /* end of list */ } }, }; But libvirt still checks for QEMU_CAPS_DRIVE_DISCARD using QMP query-command-line-options: static struct virQEMUCapsCommandLineProps virQEMUCapsCommandLine[] = { { "machine", "mem-merge", QEMU_CAPS_MEM_MERGE }, { "drive", "discard", QEMU_CAPS_DRIVE_DISCARD }, { "realtime", "mlock", QEMU_CAPS_MLOCK }, }; ... qemuMonitorGetCommandLineOptionParameters(mon, virQEMUCapsCommandLine[i].option, &values) So, when I try to use discard option in domain xml I get this error: error : qemuBuildDriveStr:3986 : unsupported configuration: discard is not supported by this QEMU binary

3 3

Re: [libvirt] Unable to provision VM attaching it directly to a OVS bridge
by Lucas Brasilino 31 Oct '13

31 Oct '13

Moving this discussion from libvirt-users: Getting deeper and deeper :) $ egrep '^int vir*' util/virnetdevopenvswitch.c int virNetDevOpenvswitchAddPort(const char *brname, const char *ifname, int virNetDevOpenvswitchRemovePort(const char *brname ATTRIBUTE_UNUSED, const char *ifname) int virNetDevOpenvswitchGetMigrateData(char **migrate, const char *ifname) int virNetDevOpenvswitchSetMigrateData(char *migrate, const char *ifname) Libvirt by now basically support adding/removing port to a OVS bridge . Is there some planning on an OVS full support, like bridge creation/destroy, creating a VM attaching it to a OVS brige, etc, just like 'traditional' Linux bridge ? If so, Is anybody working on it ? regards Lucas Brasilino MSc Student @ Federal University of Pernambuco (UFPE) twitter: @lucas_brasilino 2013/10/29 Lucas Brasilino <lrbbs(a)cin.ufpe.br>: > Hi > > Getting deeper, the error is raised by 'virDomainCreateLinux()'. > Here comes the snippet of the XML argument passed to this call: > > (Pdb) print xmlDesc > <domain type='kvm'> > <name>vm2</name> > <uuid>3d713513-e8ee-994a-0eba-51128bd4b42e</uuid> > [...] > <interface type='bridge'> > <source bridge='databr0'/> > <mac address='00:00:00:00:00:03'/> > <model type='virtio'/> > </interface> > [...] > </domain> > > So libvirt network driver is surely trying to use 'databr0' as the > common Linux bridge implementation, not an Open vSwitch bridge. > > I tried to add an > > <virtualport='openvswitch'/> > > element in the XML created by virt-install and use 'virsh create' to > create it but > seems that libvirt's network driver does not support Open vSwitch when > creating VM's. > Is there some planning to support it ? > > Thanks! > > regards > Lucas Brasilino > MSc Student @ Federal University of Pernambuco (UFPE) > twitter: @lucas_brasilino > > 2013/10/28 Lucas Brasilino <lrbbs(a)cin.ufpe.br>: >> Reposting from virt-tools mailing list: >> >> Hi! >> >> I'm facing a problem that could be triggered by some lacking >> of support from libvirt on Open vSwitch (or could be my mistake). >> >> I have interests in researching on virtual networks and >> SDN. To keep things simple, I've decided to use libvirt/virt-tools to >> manage VM's since my focus is on the network, instead of using a >> full feature system like OpenStack. >> >> I'm quite new with libvirt/virt-tools, but I have a good experience with >> openvswitch and other virtualizations technologies (which I dropped in >> libvirt/kvm favor). >> >> I'm using Fedora 19 packages >> >> openvswitch (1.11.0-1.fc19.x86_64) >> libvirt (1.0.5.6-3.fc19.x86_64) >> virt-install (0.10.0-4.fc19.noarch) >> >> I've created an OVS bridge (databr0) outside libvirt, and then defined it >> inside libvirt with: >> >> <network> >> <name>databr0</name> >> <forward mode='bridge'/> >> <bridge name='databr0'/> >> <virtualport type='openvswitch'/> >> </network> >> >> And then 'net-autostart' and 'net-start' it with virsh. Now I've got: >> # virsh net-list >> Name State Autostart Persistent >> ---------------------------------------------------------- >> databr0 active yes yes >> >> When I try to provision an VM, if I use the virt-install option >> "--network=bridge:databr0,model=virtio,mac=00:00:00:00:00:03" >> I got the following error: >> >> # virt-install --connect qemu:///system --virt-type kvm --name vm2 >> --ram 768 --disk path=/home/lucas/local/vm/images/vm2.img --vnc >> --cdrom /home/lucas/local/vm/fc19-x86_64.iso >> --network=bridge:databr0,model=virtio,mac=00:00:00:00:00:03 >> --os-type=linux --os-variant fedora19 >> >> Starting install... >> ERROR Unable to add bridge databr0 port vnet0: Operation not supported >> Domain installation does not appear to have been successful. >> If it was, you can restart your domain by running: >> virsh --connect qemu:///system start vm2 >> otherwise, please restart your installation. >> >> I just managed to create a VM when I use '--nonetworks' option and after >> I do a 'virtsh edit vm2' and add: >> >> <interface type='bridge'> >> <mac address='00:00:00:00:00:03'/> >> <source bridge='databr0'/> >> <virtualport type='openvswitch'/> >> <model type='virtio'/> >> <address type='pci' domain='0x0000' bus='0x00' slot='0x05' >> function='0x0'/> >> </interface> >> >> Well, I read elsewhere that openvswitch bridging isn't fully >> supported. Is it the case or I'm facing another kind of problem ? >> >> regards >> >> Att >> Lucas Brasilino >> MSc Student @ Federal University of Pernambuco (UFPE) >> twitter: @lucas_brasilino

2 3

[libvirt] [PATCH] LXC: mount /dev/pts/0 to /dev/console
by Gao feng 31 Oct '13

31 Oct '13

Now, /dev/console is linked to the /dev/pts/0, so for the process agetty, the tty device of agetty is pts/0. this will cause login container failed. since pts/0 is not in the /etc/securetty. so pam module pam_securetty will prevent the root user logging on the system. this patch doesn't make /dev/console a symbol but binds /dev/pts/0 to it. so the tty device of agetty will be console. root can login the system successfully. Signed-off-by: Gao feng <gaofeng(a)cn.fujitsu.com> --- src/lxc/lxc_container.c | 19 +++++++++++++------ 1 file changed, 13 insertions(+), 6 deletions(-) diff --git a/src/lxc/lxc_container.c b/src/lxc/lxc_container.c index 255c711..1cede41 100644 --- a/src/lxc/lxc_container.c +++ b/src/lxc/lxc_container.c @@ -1049,12 +1049,19 @@ static int lxcContainerSetupDevices(char **ttyPaths, size_t nttyPaths) return -1; } VIR_FREE(tty); - if (i == 0 && - symlink(ttyPaths[i], "/dev/console") < 0) { - virReportSystemError(errno, - _("Failed to symlink %s to /dev/console"), - ttyPaths[i]); - return -1; + if (i == 0) { + if (virFileTouch("/dev/console", 0600) < 0) { + virReportSystemError(errno, "%s", + _("Failed to create /dev/console")); + return -1; + } + + if (mount(ttyPaths[0], "/dev/console", NULL, MS_BIND, NULL) < 0) { + virReportSystemError(errno, + _("Failed to symlink %s to /dev/console"), + ttyPaths[i]); + return -1; + } } } return 0; -- 1.8.3.1

2 3

[libvirt] [PATCH] qemu: fix well-formed migration URI formatting
by Michael Chapman 31 Oct '13

31 Oct '13

When adding an automatically allocated port to a well-formed migration URI, keep it well-formed: tcp://1.2.3.4/ -> tcp://1.2.3.4/:12345 # wrong tcp://1.2.3.4/ -> tcp://1.2.3.4:12345/ # fixed tcp://1.2.3.4 -> tcp://1.2.3.4:12345 # still works tcp:1.2.3.4 -> tcp:1.2.3.4:12345 # still works (old syntax) Signed-off-by: Michael Chapman <mike(a)very.puzzling.org> --- src/qemu/qemu_migration.c | 16 +++++++++++++--- 1 file changed, 13 insertions(+), 3 deletions(-) diff --git a/src/qemu/qemu_migration.c b/src/qemu/qemu_migration.c index 0439ba4..cb59620 100644 --- a/src/qemu/qemu_migration.c +++ b/src/qemu/qemu_migration.c @@ -2535,6 +2535,7 @@ qemuMigrationPrepareDirect(virQEMUDriverPtr driver, char *uri_str = NULL; int ret = -1; virURIPtr uri = NULL; + bool well_formed_uri = true; VIR_DEBUG("driver=%p, dconn=%p, cookiein=%s, cookieinlen=%d, " "cookieout=%p, cookieoutlen=%p, uri_in=%s, uri_out=%p, " @@ -2597,6 +2598,7 @@ qemuMigrationPrepareDirect(virQEMUDriverPtr driver, /* Convert uri_in to well-formed URI with // after tcp: */ if (!(STRPREFIX(uri_in, "tcp://"))) { + well_formed_uri = false; if (virAsprintf(&uri_str, "tcp://%s", p) < 0) goto cleanup; } @@ -2626,9 +2628,17 @@ qemuMigrationPrepareDirect(virQEMUDriverPtr driver, goto cleanup; } - /* Caller frees */ - if (virAsprintf(uri_out, "%s:%d", uri_in, port) < 0) - goto cleanup; + if (well_formed_uri) { + uri->port = port; + + /* Caller frees */ + if (!(*uri_out = virURIFormat(uri))) + goto cleanup; + } else { + /* Caller frees */ + if (virAsprintf(uri_out, "%s:%d", uri_in, port) < 0) + goto cleanup; + } } else { port = uri->port; -- 1.8.3.1

2 3

[libvirt] [PATCH]virsh: track alias option and improve error message when option duplicate its alias
by Chen Hanxiao 31 Oct '13

31 Oct '13

From: Chen Hanxiao <chenhanxiao(a)cn.fujitsu.com> commit 2b172a8effa712aee97a21a64d2d02060958f9b2 allow alias to expand to opt=value pair. That means alias may not look alike since then. With this patch we will also track alias. If we type command with one option and another marked as its alias, we will get an error message like: error: option '--AA' duplicate its alias '--AAA' Signed-off-by: Chen Hanxiao <chenhanxiao(a)cn.fujitsu.com> --- tools/virsh.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/tools/virsh.c b/tools/virsh.c index bad78c9..423e2d8 100644 --- a/tools/virsh.c +++ b/tools/virsh.c @@ -1101,11 +1101,18 @@ vshCmddefGetOption(vshControl *ctl, const vshCmdDef *cmd, const char *name, if (VIR_STRDUP(*optstr, value + 1) < 0) goto cleanup; } + *opts_seen |= 1 << i; continue; } if ((*opts_seen & (1 << i)) && opt->type != VSH_OT_ARGV) { - vshError(ctl, _("option --%s already seen"), name); - goto cleanup; + if ((*opts_seen & (1 << (i - 1)))) { + vshError(ctl, _("option '--%s' duplicates its alias '--%s'"), + cmd->opts[i].name, cmd->opts[i-1].name); + goto cleanup; + } else { + vshError(ctl, _("option '--%s' already seen"), name); + goto cleanup; + } } *opts_seen |= 1 << i; *opt_index = i; -- 1.8.2.1

3 4

[libvirt] [PATCH] maint: avoid further typedef accidents
by Eric Blake 31 Oct '13

31 Oct '13

To make it easier to forbid future attempts at a confusing typedef name ending in Ptr that isn't actually a pointer, insist that we follow our preferred style of 'typedef foo *fooPtr'. * cfg.mk (sc_forbid_const_pointer_typedef): Enforce consistent style, to prevent issue fixed in previous storage patch. * src/conf/capabilities.h (virCapsPtr): Fix offender. * src/security/security_stack.c (virSecurityStackItemPtr): Likewise. * tests/qemucapabilitiestest.c (testQemuDataPtr): Likewise. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- cfg.mk | 4 ++++ src/conf/capabilities.h | 4 ++-- src/security/security_stack.c | 2 +- tests/qemucapabilitiestest.c | 2 +- 4 files changed, 8 insertions(+), 4 deletions(-) diff --git a/cfg.mk b/cfg.mk index e9da282..1b2fd46 100644 --- a/cfg.mk +++ b/cfg.mk @@ -471,10 +471,14 @@ sc_correct_id_types: # "const fooPtr a" is the same as "foo * const a", even though it is # usually desired to have "foo const *a". It's easier to just prevent # the confusing mix of typedef vs. const placement. +# Also requires that all 'fooPtr' typedefs are actually pointers. sc_forbid_const_pointer_typedef: @prohibit='(^|[^"])const \w*Ptr' \ halt='"const fooPtr var" does not declare what you meant' \ $(_sc_search_regexp) + @prohibit='typedef [^(]+ [^*]\w*Ptr\b' \ + halt='use correct style and type for Ptr typedefs' \ + $(_sc_search_regexp) # Forbid sizeof foo or sizeof (foo), require sizeof(foo) sc_size_of_brackets: diff --git a/src/conf/capabilities.h b/src/conf/capabilities.h index 5bc7bb5..ba99e1a 100644 --- a/src/conf/capabilities.h +++ b/src/conf/capabilities.h @@ -1,7 +1,7 @@ /* * capabilities.h: hypervisor capabilities * - * Copyright (C) 2006-2008, 2010, 2012 Red Hat, Inc. + * Copyright (C) 2006-2013 Red Hat, Inc. * Copyright (C) 2006-2008 Daniel P. Berrange * * This library is free software; you can redistribute it and/or @@ -162,7 +162,7 @@ struct _virDomainXMLNamespace { }; typedef struct _virCaps virCaps; -typedef virCaps* virCapsPtr; +typedef virCaps *virCapsPtr; struct _virCaps { virObject parent; diff --git a/src/security/security_stack.c b/src/security/security_stack.c index ff0f06b..0d42b21 100644 --- a/src/security/security_stack.c +++ b/src/security/security_stack.c @@ -30,7 +30,7 @@ typedef struct _virSecurityStackData virSecurityStackData; typedef virSecurityStackData *virSecurityStackDataPtr; typedef struct _virSecurityStackItem virSecurityStackItem; -typedef virSecurityStackItem* virSecurityStackItemPtr; +typedef virSecurityStackItem *virSecurityStackItemPtr; struct _virSecurityStackItem { virSecurityManagerPtr securityManager; diff --git a/tests/qemucapabilitiestest.c b/tests/qemucapabilitiestest.c index 28f12e7..d912171 100644 --- a/tests/qemucapabilitiestest.c +++ b/tests/qemucapabilitiestest.c @@ -27,7 +27,7 @@ #define VIR_FROM_THIS VIR_FROM_NONE typedef struct _testQemuData testQemuData; -typedef testQemuData * testQemuDataPtr; +typedef testQemuData *testQemuDataPtr; struct _testQemuData { virDomainXMLOptionPtr xmlopt; const char *base; -- 1.8.3.1

2 2

Re: [libvirt] [libvirt-users] bridged networking using VLAN : guest with 2 NIC
by Laine Stump 31 Oct '13

31 Oct '13

On 10/30/2013 09:14 PM, Dan Sa wrote: > I created another Virtual Machine with DHCP enabled 192.168.0.0/16 > <http://192.168.0.0/16> DHCP Range 192.168.0.2 192.168.0.2 (as I want > same IP to all three guests) Forwarding : NAT to em2 > > now on host I see two instances of dnsmasq > > netstat -anlp | grep -w dnsmasq > tcp 0 0 192.168.0.1:53 <http://192.168.0.1:53> > 0.0.0.0:* LISTEN 1907/dnsmasq <<<< is this right ? > I created > tcp 0 0 192.168.122.1:53 <http://192.168.122.1:53> > 0.0.0.0:* LISTEN 6666/dnsmasq <<< default NAT > > > however I am seeing dnsmasq is not Active > > systemctl status dnsmasq.service > dnsmasq.service - DNS caching server. > Loaded: loaded (/usr/lib/systemd/system/dnsmasq.service; > enabled) > Active: failed (Result: exit-code) since Wed, 30 Oct 2013 > 12:30:39 -0600; 41min ago > > all my guest are on physical VLAN so how I should go about giving them > 192.168.0.2 address using dnsmasq ? 1) PLEASE stop top-posting new random information and guesses at solutions to the top of your old messages. You are quickly creating a cascade of text that is so large and rambling that nobody will have the ambition to unscramble it and make a reply. It is generally accepted good etiquette to post responses to earlier messages in-line with the original, so that the flow of the conversation can be easily followed by someone reading just the last message. (If there is no "flow" to the conversation, then something more fundamental has gone wrong.) 2) If you've now realized (as I told you in my first response) that you need to run a system instance of dhcpd or dnsmasq to service dhcp for the vlan of your guests, then what you have is a networking problem, not a virtualization problem. There are tutorials for setting up dnsmasq in various places on the web, or you can simply read the comments in /etc/dnsmasq.conf to figure out how to make it listen on just a specific interface, and how to tell it what addresses to serve for that interface. See below for more comments. > > thanks > dan > > > > > > > On Wed, Oct 30, 2013 at 11:20 AM, Dan Sa <dreamrider787(a)gmail.com > <mailto:dreamrider787@gmail.com>> wrote: > > UPDATE : > > Adding more information about dnsmasq on box running fedora (host) > > dispatcher.d]# nmcli nm status > RUNNING STATE WIFI-HARDWARE WIFI > WWAN-HARDWARE WWAN > running connecting enabled enabled > enabled disabled > > nmcli nm status > RUNNING STATE WIFI-HARDWARE WIFI > WWAN-HARDWARE WWAN > running disconnected enabled enabled > enabled disabled > > > systemctl --all|grep -i network > > NetworkManager.service loaded active running Network > Manager > > dnsmasq.d]# ls -al > total 8 > drwxr-xr-x 2 root root 4096 Nov 20 2012 . > drwxr-xr-x. 6 root root 4096 Oct 30 08:56 .. > > > dispatcher.d]# ls -al > total 32 > drwxr-xr-x. 2 root root 4096 Sep 23 12:59 . > drwxr-xr-x. 6 root root 4096 Oct 30 08:56 .. > -rwxr-xr-x 1 root root 175 Nov 15 2012 00-netreport > -rwxr-xr-x. 1 root root 335 Feb 14 2012 04-iscsi > -rwxr-xr-x. 1 root root 111 Jun 25 2012 10-sendmail > -rwxr-xr-x 1 root root 933 Jul 10 04:38 11-dhclient > -rwxr-xr-x. 1 root root 317 Apr 27 2012 20-chrony > -rwxr-xr-x 1 root root 138 Mar 19 2013 20-squid > > cat dnsmasq.conf > > all lines are commented only line active is > > # Include another lot of configuration options. > #conf-file=/etc/dnsmasq.more.conf > conf-dir=/etc/dnsmasq.d <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< > *All* of the above is just random data, which says nothing. Please try to spend more time figuring out what you are trying to do and what has gone wrong before just pasting random bits into a message to a mailing list. Remember that the people reading the messages on this list are doing it for free, and they also have day jobs they need to attend to. The less excess garbage is in a message, the less time they have to use up sifting through it for useful information. > > > > > > On Wed, Oct 30, 2013 at 10:10 AM, Dan Sa <dreamrider787(a)gmail.com > <mailto:dreamrider787@gmail.com>> wrote: > > BRCTL SHOW > bridge name bridge id STP enabled interfaces > br0 8000.00237de0a132 no em1 > vnet0 > guest1-lan 8000.00237de0a133 no em2.620 > vnet3 > guest1-wan 8000.0022642b6586 no p1p1.110 > guest2-lan 8000.00237de0a133 no em2.619 > guest2-wan 8000.0022642b6586 no p1p1.108 > guest3-lan 8000.00237de0a133 no em2.621 > guest3-wan 8000.0022642b6586 no p1p1.111 > > virbr0 8000.5254003e19b3 yes virbr0-nic > vnet2 > The above list of bridges connected to vlan interfaces implies that, for some reason, you are putting each interface of each guest on a separate vlan. Is this really what you think you want to do? Why? How will these guests connect to the rest of the network? Unless you have a requirement to isolate each guest entirely from every other guest and force each guest's traffic to travel through a different path out to the rest of the network, then I think you've misunderstood when/why vlans should be used. It is much more common to just setup a single network that many (or even all) guests will attach to, and generally you don't need a vlan tag to do this. (an earlier comment you made implying that a *host* bridge attached to a host vlan interface *was actually the guest interface* makes me believe you are confused about the function of each bit and how they all fit together. I think it would be useful for you to step back from all of this and just explain how you want your guests to connect to the rest of the network. Do you have a specific requirement for each to be connected to a separate vlan on the physical network? Or do they just need to be connected "somehow"? Do they need to be prevented from communicating with each other, or was this just a side effect of a mistaken network configuration? > > > On Wed, Oct 30, 2013 at 10:10 AM, Dan Sa > <dreamrider787(a)gmail.com <mailto:dreamrider787@gmail.com>> wrote: > > Laine, > > first thank you very much for your input, it cleared some > doubts ... please find below the information in the > desired format. > > Versions: > > 1) Virtual Machine Manager 0.9.5[root@dsl-test qemu]# uname -a > 2) Linux 3.9.10-100.fc17.x86_64 #1 SMP Sun Jul 14 > 01:31:27 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux > 3) Fedora release 17 (Beefy Miracle) Kernel \r on an \m (\l) > Fedora 17 is > 1 year old, and thus no longer gets any updates, not even security updates. You need to upgrade to a newer release of Fedora (F20 is already in beta, so that may be a good choice) > 4) virsh --version 0.9.11.10 > Likewise, 0.9.11 is very old. *Many* improvements have been made since then, which is another good reason to upgrade your OS. > 5) rpm -qa | grep kvm > libvirt-daemon-kvm-0.9.11.10-1.fc17.x86_64 > qemu-kvm-1.0.1-6.fc17.x86_64 > > I have 3 Guest information for guest1 follows : (my > questions are after that) > > virsh dumpxml guest1 : > > virsh dumpxml guest1 > > <domain type='kvm' id='5'> > > <name>guest1</name> > > <uuid>bcc29de1-ba90-5dd8-81fe-7450852df7f6</uuid> > > <memory unit='KiB'>1048576</memory> > > <currentMemory unit='KiB'>1048576</currentMemory> > > <vcpu>4</vcpu> > > <os> > > <type arch='x86_64' machine='pc-0.15'>hvm</type> > > <boot dev='hd'/> > > </os> > > <features> > > <acpi/> > > <apic/> > > <pae/> > > </features> > > <cpu mode='custom' match='exact'> > > <model fallback='allow'>Penryn</model> > > <vendor>Intel</vendor> > > <feature policy='require' name='tm2'/> > > <feature policy='require' name='est'/> > > <feature policy='require' name='monitor'/> > > <feature policy='require' name='osxsave'/> > > <feature policy='require' name='ss'/> > > <feature policy='require' name='ds'/> > > <feature policy='require' name='vme'/> > > <feature policy='require' name='dtes64'/> > > <feature policy='require' name='ht'/> > > <feature policy='require' name='dca'/> > > <feature policy='require' name='pbe'/> > > <feature policy='require' name='xsave'/> > > <feature policy='require' name='tm'/> > > <feature policy='require' name='pdcm'/> > > <feature policy='require' name='vmx'/> > > <feature policy='require' name='ds_cpl'/> > > <feature policy='require' name='xtpr'/> > > <feature policy='require' name='acpi'/> > > </cpu> > > <clock offset='utc'/> > > <on_poweroff>destroy</on_poweroff> > > <on_reboot>restart</on_reboot> > > <on_crash>restart</on_crash> > > <devices> > > <emulator>/usr/bin/qemu-kvm</emulator> > > <disk type='file' device='disk'> > > <driver name='qemu' type='raw'/> > > <source file='/var/lib/libvirt/images/guest1.img'/> > > <target dev='vda' bus='virtio'/> > > <alias name='virtio-disk0'/> > > <address type='pci' domain='0x0000' bus='0x00' > slot='0x05' function='0x0'/> > > </disk> > > <controller type='usb' index='0'> > > <alias name='usb0'/> > > <address type='pci' domain='0x0000' bus='0x00' > slot='0x01' function='0x2'/> > > </controller> > > <controller type='virtio-serial' index='0'> > > <alias name='virtio-serial0'/> > > <address type='pci' domain='0x0000' bus='0x00' > slot='0x04' function='0x0'/> > > </controller> > > <interface type='network'> > > <mac address='52:54:00:ea:80:df'/> > > <source network='default'/> > > <target dev='vnet2'/> > > <alias name='net0'/> > > <address type='pci' domain='0x0000' bus='0x00' > slot='0x03' function='0x0'/> > > </interface> > > <interface type='bridge'> > > <mac address='52:54:00:d5:1d:01'/> > > <source bridge='guest1-lan'/> > > <target dev='vnet3'/> > > <model type='virtio'/> > > <alias name='net1'/> > > <address type='pci' domain='0x0000' bus='0x00' > slot='0x07' function='0x0'/> > > </interface> > > <serial type='pty'> > > <source path='/dev/pts/3'/> > > <target port='0'/> > > <alias name='serial0'/> > > </serial> > > <console type='pty' tty='/dev/pts/3'> > > <source path='/dev/pts/3'/> > > <target type='serial' port='0'/> > > <alias name='serial0'/> > > </console> > > <input type='mouse' bus='ps2'/> > > <graphics type='vnc' port='5901' autoport='yes'/> > > <video> > > <model type='cirrus' vram='9216' heads='1'/> > > <alias name='video0'/> > > <address type='pci' domain='0x0000' bus='0x00' > slot='0x02' function='0x0'/> > > </video> > > <memballoon model='virtio'> > > <alias name='balloon0'/> > > <address type='pci' domain='0x0000' bus='0x00' > slot='0x06' function='0x0'/> > > </memballoon> > > </devices> > > <seclabel type='none'/> > > </domain> > > > > > > Here is revised BRCTL Show after guest1 is running > > > BRCTL SHOW > > bridge name bridge id STP enabled interfaces > > br0 8000.00237de0a132 > no em1 > > vnet0 > > c1000z-lan 8000.00237de0a133 > no em2.620 > > vnet3 > > c1000z-wan 8000.0022642b6586 > no p1p1.110 > > c2000t-lan 8000.00237de0a133 > no em2.619 > > c2000t-wan 8000.0022642b6586 no > p1p1.108 > > pk5001a-lan 8000.00237de0a133 > no em2.621 > > pk5001a-wan 8000.0022642b6586 > no p1p1.111 > > virbr0 8000.5254003e19b3 yes > virbr0-nic > > vnet2 > The above is obviously *not* the output of brctl on the same host as the above-listed guest, since that guest connects to a bridge called "guest1-lan", which isn't even in the brctl output at all. > ---------------------------------------------------------------------------------------------------------- > > > on guest1-lan terminal I see following : > > > route -n > > destination gateway genmask flags > metric ref use iface > > 0.0.0.0 192.168.122.1 0.0.0.0 UG > 100 0 eth0 > > 192.168.122.0 0.0.0.0 255.255.255.0 U > 0 0 eth0 > > > DHCPDISCOVER on eth 1 to 255.255.255.255 port 67 interval > 6 then 14, 13, 15 > > and then sleeping > > > so you are right I am not getting DHCP IP and I have > "forward mode bridge" so libvirt is ignoring it > > I want 192.168.0.2 on this with gateway 192.168.0.1 > > > so looks like my problem is DHCP .... how can I able to > run DHCP for all three guest which are part of em2 > > interface like em2.629, em2.621 and em2/622 > Again, WHY do you need a separate vlan for each guest? I think you may be confused. > > awaiting your suggestions.... > 1) stop top-posting. 2) either explain why you need each guest on a separate vlan, or give up on that idea and just connect all guest interfaces to a single network (or possible 2 networks - on lan and one wan, which you seen to have some requirement for). 3) if you really do require the guests to each be on a separate vlan, and you don't have any other host on the network that already serves dhcp on those vlans, then read /etc/dnsmasq.conf to learn how to configure it. > > regards > > dan > > > > On Wed, Oct 30, 2013 at 9:56 AM, Dan Sa > <dreamrider787(a)gmail.com <mailto:dreamrider787@gmail.com>> > wrote: > > Laine, > > first thank you very much for your input, it cleared > some doubts ... please find below the information in > the desired format. > > Versions: > > 1) Virtual Machine Manager 0.9.5[root@dsl-test qemu]# > uname -a > 2) Linux 3.9.10-100.fc17.x86_64 #1 SMP Sun Jul 14 > 01:31:27 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux > 3) Fedora release 17 (Beefy Miracle) Kernel \r on an > \m (\l) > 4) virsh --version 0.9.11.10 > 5) rpm -qa | grep kvm > libvirt-daemon-kvm-0.9.11.10-1.fc17.x86_64 > qemu-kvm-1.0.1-6.fc17.x86_64 > > I have 3 Guest and here is > > > > > On Wed, Oct 30, 2013 at 5:19 AM, Laine Stump > <laine(a)laine.org <mailto:laine@laine.org>> wrote: > > (There is no need or advantage to Cc'ing > individuals who are already > subscribed to the mailing list.) > > On 10/28/2013 05:34 PM, Dan Sa wrote: > > hello all, > > > > I have been trying to set-up bridged network > with VLAN and not able to > > succeed as many tutorials address only single NIC. > > > > I am trying to setup 2 guests (backtrack > instance) each guest has NIC1 > > and NIC2. following is snippet for guest1 > > > > I am not able to get 192.168.0.2 address back on > guest eth0. > > > See the comment below about <forward > mode='bridge'>. you'll need some > other entity on your vlan to run a dhcp server, > because libvirt won't be > doing it for you in this case. > > > > > > > > VIRT-MANAGER GUI : > > > > guest1-lan details radio button > > > > left side panel > > > > NIC1 ------------------> Virtual Network Interface > > Source Device : > Virtual Network "default" NAT > > Device Model : > Hypervisor default > > MAc Address : > xxxxxxxxxxxxxx > > > > NIC2 ------------------> Virtual Network Interface > > Source Device : > Specify Shared Device Name > > Bridge > name : guest1-lan > > Device Model : virto > > MAc Address : > xxxxxxxxxxxxxx > > The output of "virsh dumpxml $guestname" is much > more useful than a > transcription of the virt-manager screens. > > > > > HOST MACHINE : > > > > brctl show has br0 for bridge > > and virbr0 with 192.168.122.x address (created > by default virtual > > network NAT) > > > > > > /etc/sysconfig/network-scripts/ > > > > 1) Bridge BR0 (cat ifcfg-br0) > > > > DEVICE="br0" > > TYPE="Bridge" > > ONBOOT="yes" > > NM_CONTROLLED="no" > > BOOTPROTO="static" > > IPADDR="xx.xx.xx.xx" > > NETMASK="255.255.254.0" > > GATEWAY="xx.xx.xx.xx" > > DNS1="x.y.z.s" > > DNS2="x.y.q.s" > > > > 2) cat ifcfg-em1 > > NM_CONTROLLED="yes" > > HWADDR="02:12D:E2:B1:32" > > BOOTPROTO="static" > > DEVICE="em1" > > BRIDGE="br0" > > ONBOOT="yes" > > > > 3) ifcfg-em2 > > NM_CONTROLLED="yes" > > HWADDR="02:24:7e:d0:b1:42" > > BOOTPROTO="static" > > DEVICE="em2" > > ONBOOT="yes" > > > > 4) THIS IS GUEST (cat ifcfg-guest1-lan) > > I don't understand what you mean by "this is > guest". It isn't a part of > the guest; it is a bridge on the host that could > be *used* by a guest. > > > > DEVICE=guest1-lan > > TYPE=Bridge > > ONBOOT=yes > > BOOTPROTO=static > > DELAY=1 > > > > 5) GUEST VLAN (cat ifcfg-em2.620) > > > > DEVICE=em2.620 > > VLAN=yes > > ONBOOT=yes > > BRIDGE=guest1-lan > > > > BRCTL Show Command : > > > > br0 8000.00237de0a132 no em1 > > > vnet0 > > guest1-lan 8000.00237de0a133 no em2.620 > > virbr0 8000.5254003e19b3 yes virbr0-nic > > > From the above, it appears that there is only a > single guest running, > and that it is connected via the br0 bridge; > apparently you took this > output when neither of your dual-nic guests were > running, as they should > have each attached tun devices to both guest1-lan > and virbr0. > > > > > > > > VIRSH : > > > > virsh # net-list > > Name State Autostart > > ----------------------------------------- > > guest1-lan active yes > > default active yes > > > > > > virsh # iface-list > > Name State MAC Address > > -------------------------------------------- > > br0 active 00:23:7d:e0:a1:32 > > guest1-lan active 00:23:7d:e0:a1:33 > > > > > > iface-edit : > > > > virsh # iface-edit guest1-lan > > > > <interface type='bridge' name='guest1-lan'> > > <start mode='onboot'/> > > <bridge delay='1'> > > <interface type='vlan' name='em2.620'> > > <vlan tag='620'> > > <interface name='em2'/> > > </vlan> > > </interface> > > </bridge> > > </interface> > > > > > ------------------------------------------------------------------ > > > > /etc/libvirt/qemu/networks > > > (You shouldn't be looking at/modifying the files in > /etc/libvirt/qemu/networks directly. Instead, use > "virsh net-dumpxml > guest1-lan" (for example) to look at the network > config, and "virsh > net-edit guest1-lan" to modify it.) > > > > > > cat guest1-lan.xml > > <network> > > <name>guest1-lan</name> > > <uuid>a12747ec-21c9-0d21-ab06-064ba204bc52</uuid> > > <forward mode='bridge' dev="br0"/> > > <bridge name='guest1-lan' /> > > <ip address='192.168.0.1' netmask='255.255.255.0'> > > <dhcp> > > <range start='192.168.0.2' > end='192.168.0.254' /> > > </dhcp> > > </ip> > > > Any network with <forward mode='bridge'...> is an > "unmanaged" network > from libvirt's POV, and thus the <ip> element and > all its subelements > are ignored. If you use <forward mode='bridge'> > then libvirt assumes > that the bridge device is already configured by > the base OS config. > > As of libvirt-1.0.1, attempts to define an <ip> > element in a network > with <forward mode='bridge'> are flagged as an > error. (It would be > helpful in future reports if you indicate your 1) > libvirt version, 2) > qemu version, 3) distro and version, 4) kernel > version. Although not > always applicable, sometime it can help in framing > the issue. > > > > </network> > > > > > > cat default.xml > > > > <network> > > <name>default</name> > > <uuid>8778244b-1a0c-c15f-c348-26462a07a639</uuid> > > <forward mode='nat'/> > > <bridge name='virbr0' stp='on' delay='0' /> > > <mac address='52:54:00:3E:19:B3'/> > > <ip address='192.168.122.1' > netmask='255.255.255.0'> > > <dhcp> > > <range start='192.168.122.2' > end='192.168.122.254' /> > > </dhcp> > > </ip> > > </network> > > > > any guidance will be appriciated > > Since you're defining a vlan tag, I assume that > the physical network > attached to your host's em2 is actually using vlan > 620? If not, and you > just need a network that's private to your guests > and the host, I would > recommend simply defining a libvirt network with > no <forward> element at > all. This network *will* be managed by libvirt, so > libvirt will create a > bridge and give it an IP address, as well as > running a dnsmasq instance > to serve up IP addresses to guests, but the guests > won't be able to get > traffic anywhere beyond that bridge via their > interface connected to the > bridge. > > If you *are* using vlan 620 on the physical > network, then you'll need to > setup some other dhcp server somewhere on that > network (either run a > system instance of dnsmasq on the host that > listens on em2.620, or run > dnsmasq or dhcpd on some other physical host or > guest that listens on > its own vlan-tagged interface). > > > > > >

1 0

[libvirt] RFC: using a network device as a destination for a disk snapshot
by Eric Blake 31 Oct '13

31 Oct '13

Right now, when creating an external snapshot, the snapshot xml requires that the destination be in the local file system: http://libvirt.org/formatsnapshot.html <domainsnapshot> ... <disks> <disk name='vda' snapshot='external'> <driver type='qcow2'/> <source file='/path/to/new'/> </disk> ... But libvirt already allows for a network device with qcow2 contents, provided that there is no backing chain: # qemu-img info gluster://red/vol1/img2 image: gluster://red/vol1/img2 file format: qcow2 virtual size: 10M (10485760 bytes) disk size: 193K cluster_size: 65536 # virsh dumpxml dom <domain type='kvm'> ... <disk type='network' device='disk'> <driver name='qemu' type='qcow2'/> <source protocol='gluster' name='vol1/img2'> <host name='red'/> </source> <target dev='vdc' bus='virtio'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x09' function='0x0'/> </disk> ... I'm working on patches to the backing chain code to allow a gluster (or other network device) with qcow2 contents to have a backing file, but for the code to be useful, we also need to patch <domainsnapshot> xml to allow the destination file to be a gluster or other network device, rather than forcing it to be a local file name. Here's the XML I think we need to add to domainsnapshot: <domainsnapshot> ... <disks> <disk name='vda' snapshot='external' type='network'> <driver type='qcow2'/> <source protocol='gluster' name='vol1/img2'> <host name='red'/> </source> </disk> that is, add an optional /disk@type attribute (if absent, it defaults to type='file'), and where if present, the <source> subelement then takes on alternate forms in the same manner in which //domain/devices/disk handles alternates (here, allowing a protocol, name, and host specification). [Ultimately, we need to fix //domain/devices/disk to specify a full backing chain, but one step at a time...] -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org

3 2