- Devel - libvirt List Archives

[libvirt] fix "make distcheck" failures
by Jim Meyering 22 Jul '09

22 Jul '09

Running "make check" from a tarball would show two test failures, since some test-related files were not being distributed. [that is one of the things that "make distcheck" tests for] With these two fixes, "make distcheck" passes once again: >From 553cecaf27e86755e40bab346d4014fb0291e5cd Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Wed, 22 Jul 2009 05:07:07 -0400 Subject: [PATCH 1/2] avoid a make distcheck failure: distribute tests/interfaceschemadata/ * tests/Makefile.am (EXTRA_DIST): Add interfaceschemadata. --- tests/Makefile.am | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/tests/Makefile.am b/tests/Makefile.am index efd7c6e..74e98d1 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -53,6 +53,7 @@ EXTRA_DIST = \ networkschematest \ domainschematest \ domainschemadata \ + interfaceschemadata \ storagepoolschematest \ storagepoolschemadata \ storagevolschematest \ -- 1.6.2.5 >From 76f208cc5235023c594e84783af7a33383c8f649 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Wed, 22 Jul 2009 05:10:14 -0400 Subject: [PATCH 2/2] avoid a make distcheck failure: distribute docs/schemas/interface.rng * docs/schemas/Makefile.am (schema_DATA): Add interface.rng. --- docs/schemas/Makefile.am | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/docs/schemas/Makefile.am b/docs/schemas/Makefile.am index c4ef0b1..ef41a63 100644 --- a/docs/schemas/Makefile.am +++ b/docs/schemas/Makefile.am @@ -3,6 +3,7 @@ schemadir = $(pkgdatadir)/schemas schema_DATA = \ domain.rng \ + interface.rng \ network.rng \ storagepool.rng \ storagevol.rng \ -- 1.6.2.5

2 1

[libvirt] [PATCH] libvirt-glib: import gtk instead of pygtk
by Guido Günther 22 Jul '09

22 Jul '09

Hi, the python example imports pygtk but uses gtk.main() so it fails to work. Attached patch fixes this. Cheers, -- Guido

1 0

[libvirt] c&p error in api doc
by Thomas Treutner 22 Jul '09

22 Jul '09

I think there's a c&p error: http://libvirt.org/html/libvirt-libvirt.html#virStoragePoolUndefine "Returns: a virStoragePoolPtr object, or NULL if creation failed" I assume it is "0 on success, -1 on failure" as the return is of type int. kr,t

3 4

[libvirt] Problem with libvirt and pci passthrough
by Mirko Raasch 21 Jul '09

21 Jul '09

Hi, i´m still having problem with libvirt and pci passthrough, if i use more then one pci card in the guest. If i start qemu 0.10.5 manualy, i can use all of the pci devices in the guest, here is the log: http://nopaste.info/610fa81fb5.html If i use libvirt to start the guest and passthrough three pci device, none of the device is usable in the guest: http://nopaste.info/ccfbfcf9a0.html Is there any limitation in libvirt? Greetings Mirko

2 6

[libvirt] [PATCH] Make tcp_chr_read() use recvmsg()
by Mark McLoughlin 21 Jul '09

21 Jul '09

Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> --- qemu-char.c | 12 +++++++++++- 1 files changed, 11 insertions(+), 1 deletions(-) diff --git a/qemu-char.c b/qemu-char.c index 287e0cd..e0d7220 100644 --- a/qemu-char.c +++ b/qemu-char.c @@ -1911,15 +1911,25 @@ static void tcp_chr_read(void *opaque) { CharDriverState *chr = opaque; TCPCharDriver *s = chr->opaque; + struct msghdr msg = { 0, }; + struct iovec iov[1]; uint8_t buf[1024]; int len, size; if (!s->connected || s->max_size <= 0) return; + len = sizeof(buf); if (len > s->max_size) len = s->max_size; - size = recv(s->fd, (void *)buf, len, 0); + + iov[0].iov_base = buf; + iov[0].iov_len = len; + + msg.msg_iov = iov; + msg.msg_iovlen = 1; + + size = recvmsg(s->fd, &msg, 0); if (size == 0) { /* connection closed */ s->connected = 0; -- 1.6.2.5

1 1

[libvirt] [PATCH] Add support for fd=name to tap and socket networking
by Mark McLoughlin 21 Jul '09

21 Jul '09

This allows a program to initialize a host networking device using a file descriptor passed over a unix monitor socket. The program must first pass the file descriptor using SCM_RIGHTS ancillary data with the getfd monitor command. It then may do "host_net_add tap fd=name" to use the named file descriptor. Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> --- net.c | 42 +++++++++++++++++++++++++++++++++++------- 1 files changed, 35 insertions(+), 7 deletions(-) diff --git a/net.c b/net.c index 90cf912..1838692 100644 --- a/net.c +++ b/net.c @@ -2410,6 +2410,23 @@ void qemu_check_nic_model_list(NICInfo *nd, const char * const *models, exit(exit_status); } +static int net_handle_fd_param(Monitor *mon, const char *param) +{ + if (!qemu_isdigit(param[0])) { + int fd; + + fd = monitor_get_fd(mon, param); + if (fd == -1) { + config_error(mon, "No file descriptor named %s found", param); + return -1; + } + + return fd; + } else { + return strtol(param, NULL, 0); + } +} + int net_client_init(Monitor *mon, const char *device, const char *p) { char buf[1024]; @@ -2650,14 +2667,20 @@ int net_client_init(Monitor *mon, const char *device, const char *p) static const char * const fd_params[] = { "vlan", "name", "fd", "sndbuf", NULL }; + ret = -1; if (check_params(chkbuf, sizeof(chkbuf), fd_params, p) < 0) { config_error(mon, "invalid parameter '%s' in '%s'\n", chkbuf, p); - ret = -1; goto out; } - fd = strtol(buf, NULL, 0); + fd = net_handle_fd_param(mon, buf); + if (fd == -1) { + goto out; + } fcntl(fd, F_SETFL, O_NONBLOCK); s = net_tap_fd_init(vlan, device, name, fd); + if (!s) { + close(fd); + } } else { static const char * const tap_params[] = { "vlan", "name", "ifname", "script", "downscript", "sndbuf", NULL @@ -2697,15 +2720,20 @@ int net_client_init(Monitor *mon, const char *device, const char *p) "vlan", "name", "fd", NULL }; int fd; + ret = -1; if (check_params(chkbuf, sizeof(chkbuf), fd_params, p) < 0) { config_error(mon, "invalid parameter '%s' in '%s'\n", chkbuf, p); - ret = -1; goto out; } - fd = strtol(buf, NULL, 0); - ret = -1; - if (net_socket_fd_init(vlan, device, name, fd, 1)) - ret = 0; + fd = net_handle_fd_param(mon, buf); + if (fd == -1) { + goto out; + } + if (!net_socket_fd_init(vlan, device, name, fd, 1)) { + close(fd); + goto out; + } + ret = 0; } else if (get_param_value(buf, sizeof(buf), "listen", p) > 0) { static const char * const listen_params[] = { "vlan", "name", "listen", NULL -- 1.6.2.5

1 0

[libvirt] [PATCH] Add monitor_get_fd() command for fetching named fds
by Mark McLoughlin 21 Jul '09

21 Jul '09

Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> --- monitor.c | 24 ++++++++++++++++++++++++ monitor.h | 2 ++ 2 files changed, 26 insertions(+), 0 deletions(-) diff --git a/monitor.c b/monitor.c index 817e4b7..6ad2e14 100644 --- a/monitor.c +++ b/monitor.c @@ -1774,6 +1774,30 @@ static void do_closefd(Monitor *mon, const char *fdname) fdname); } +int monitor_get_fd(Monitor *mon, const char *fdname) +{ + mon_fd_t *monfd; + + LIST_FOREACH(monfd, &mon->fds, next) { + int fd; + + if (strcmp(monfd->name, fdname) != 0) { + continue; + } + + fd = monfd->fd; + + /* caller takes ownership of fd */ + LIST_REMOVE(monfd, next); + qemu_free(monfd->name); + qemu_free(monfd); + + return fd; + } + + return -1; +} + static const mon_cmd_t mon_cmds[] = { #include "qemu-monitor.h" { NULL, NULL, }, diff --git a/monitor.h b/monitor.h index 13e8cc7..f6a43c0 100644 --- a/monitor.h +++ b/monitor.h @@ -20,6 +20,8 @@ void monitor_read_bdrv_key_start(Monitor *mon, BlockDriverState *bs, BlockDriverCompletionFunc *completion_cb, void *opaque); +int monitor_get_fd(Monitor *mon, const char *fdname); + void monitor_vprintf(Monitor *mon, const char *fmt, va_list ap); void monitor_printf(Monitor *mon, const char *fmt, ...) __attribute__ ((__format__ (__printf__, 2, 3))); -- 1.6.2.5

1 0

[libvirt] [PATCH] Add getfd and closefd monitor commands
by Mark McLoughlin 21 Jul '09

21 Jul '09

Add monitor commands to support passing file descriptors via SCM_RIGHTS. getfd assigns the passed file descriptor a name for use with other monitor commands. closefd allows passed file descriptors to be closed. If a monitor command actually uses a named file descriptor, closefd will not be required. Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> --- monitor.c | 69 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ qemu-monitor.hx | 18 ++++++++++++++ 2 files changed, 87 insertions(+), 0 deletions(-) diff --git a/monitor.c b/monitor.c index e38c86e..817e4b7 100644 --- a/monitor.c +++ b/monitor.c @@ -70,6 +70,14 @@ typedef struct mon_cmd_t { const char *help; } mon_cmd_t; +/* file descriptors passed via SCM_RIGHTS */ +typedef struct mon_fd_t mon_fd_t; +struct mon_fd_t { + char *name; + int fd; + LIST_ENTRY(mon_fd_t) next; +}; + struct Monitor { CharDriverState *chr; int flags; @@ -80,6 +88,7 @@ struct Monitor { CPUState *mon_cpu; BlockDriverCompletionFunc *password_completion_cb; void *password_opaque; + LIST_HEAD(,mon_fd_t) fds; LIST_ENTRY(Monitor) entry; }; @@ -1705,6 +1714,66 @@ static void do_inject_mce(Monitor *mon, } #endif +static void do_getfd(Monitor *mon, const char *fdname) +{ + mon_fd_t *monfd; + int fd; + + fd = qemu_chr_get_msgfd(mon->chr); + if (fd == -1) { + monitor_printf(mon, "getfd: no file descriptor supplied via SCM_RIGHTS\n"); + return; + } + + if (qemu_isdigit(fdname[0])) { + monitor_printf(mon, "getfd: monitor names may not begin with a number\n"); + return; + } + + fd = dup(fd); + if (fd == -1) { + monitor_printf(mon, "Failed to dup() file descriptor: %s\n", + strerror(errno)); + return; + } + + LIST_FOREACH(monfd, &mon->fds, next) { + if (strcmp(monfd->name, fdname) != 0) { + continue; + } + + close(monfd->fd); + monfd->fd = fd; + return; + } + + monfd = qemu_mallocz(sizeof(mon_fd_t)); + monfd->name = qemu_strdup(fdname); + monfd->fd = fd; + + LIST_INSERT_HEAD(&mon->fds, monfd, next); +} + +static void do_closefd(Monitor *mon, const char *fdname) +{ + mon_fd_t *monfd; + + LIST_FOREACH(monfd, &mon->fds, next) { + if (strcmp(monfd->name, fdname) != 0) { + continue; + } + + LIST_REMOVE(monfd, next); + close(monfd->fd); + qemu_free(monfd->name); + qemu_free(monfd); + return; + } + + monitor_printf(mon, "Failed to find file descriptor named %s\n", + fdname); +} + static const mon_cmd_t mon_cmds[] = { #include "qemu-monitor.h" { NULL, NULL, }, diff --git a/qemu-monitor.hx b/qemu-monitor.hx index 70e2475..11bdb2c 100644 --- a/qemu-monitor.hx +++ b/qemu-monitor.hx @@ -628,6 +628,24 @@ STEXI Inject an MCE on the given CPU (x86 only). ETEXI + { "getfd", "s", do_getfd, "getfd name", + "receive a file descriptor via SCM rights and assign it a name" }, +STEXI +@item getfd @var{fdname} +If a file descriptor is passed alongside this command using the SCM_RIGHTS +mechanism on unix sockets, it is stored using the name @var{fdname} for +later use by other monitor commands. +ETEXI + + { "closefd", "s", do_closefd, "closefd name", + "close a file descriptor previously passed via SCM rights" }, +STEXI +@item closefd @var{fdname} +Close the file descriptor previously assigned to @var{fdname} using the +@code{getfd} command. This is only needed if the file descriptor was never +used by another monitor command. +ETEXI + STEXI @end table ETEXI -- 1.6.2.5

1 0

[libvirt] [PATCH] Add SCM_RIGHTS support to unix socket character devices
by Mark McLoughlin 21 Jul '09

21 Jul '09

If a file descriptor is passed via a message with SCM_RIGHTS ancillary data on a unix socket, store the file descriptor for use in the chr_read() handler. Close the file descriptor if it was not used. The qemu_chr_get_msgfd() API provides access to the passed descriptor. Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> --- qemu-char.c | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++ qemu-char.h | 2 ++ 2 files changed, 52 insertions(+), 0 deletions(-) diff --git a/qemu-char.c b/qemu-char.c index e0d7220..f06a614 100644 --- a/qemu-char.c +++ b/qemu-char.c @@ -168,6 +168,11 @@ void qemu_chr_read(CharDriverState *s, uint8_t *buf, int len) s->chr_read(s->handler_opaque, buf, len); } +int qemu_chr_get_msgfd(CharDriverState *s) +{ + return s->get_msgfd ? s->get_msgfd(s) : -1; +} + void qemu_chr_accept_input(CharDriverState *s) { if (s->chr_accept_input) @@ -1832,6 +1837,7 @@ typedef struct { int do_telnetopt; int do_nodelay; int is_unix; + int msgfd; } TCPCharDriver; static void tcp_chr_accept(void *opaque); @@ -1907,12 +1913,46 @@ static void tcp_chr_process_IAC_bytes(CharDriverState *chr, *size = j; } +static int tcp_get_msgfd(CharDriverState *chr) +{ + TCPCharDriver *s = chr->opaque; + + return s->msgfd; +} + +static void unix_process_msgfd(CharDriverState *chr, struct msghdr *msg) +{ + TCPCharDriver *s = chr->opaque; + struct cmsghdr *cmsg; + + for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) { + int fd; + + if (cmsg->cmsg_len != CMSG_LEN(sizeof(int)) || + cmsg->cmsg_level != SOL_SOCKET || + cmsg->cmsg_type != SCM_RIGHTS) + continue; + + fd = *((int *)CMSG_DATA(cmsg)); + if (fd < 0) + continue; + + if (s->msgfd != -1) + close(s->msgfd); + s->msgfd = fd; + } +} + static void tcp_chr_read(void *opaque) { CharDriverState *chr = opaque; TCPCharDriver *s = chr->opaque; struct msghdr msg = { 0, }; struct iovec iov[1]; + union { + struct cmsghdr cmsg; + char control[CMSG_SPACE(sizeof(int))]; + } msg_control; uint8_t buf[1024]; int len, size; @@ -1928,6 +1968,8 @@ static void tcp_chr_read(void *opaque) msg.msg_iov = iov; msg.msg_iovlen = 1; + msg.msg_control = &msg_control; + msg.msg_controllen = sizeof(msg_control); size = recvmsg(s->fd, &msg, 0); if (size == 0) { @@ -1940,10 +1982,16 @@ static void tcp_chr_read(void *opaque) closesocket(s->fd); s->fd = -1; } else if (size > 0) { + if (s->is_unix) + unix_process_msgfd(chr, &msg); if (s->do_telnetopt) tcp_chr_process_IAC_bytes(chr, s, buf, &size); if (size > 0) qemu_chr_read(chr, buf, size); + if (s->msgfd != -1) { + close(s->msgfd); + s->msgfd = -1; + } } } @@ -2105,12 +2153,14 @@ static CharDriverState *qemu_chr_open_tcp(const char *host_str, s->connected = 0; s->fd = -1; s->listen_fd = -1; + s->msgfd = -1; s->is_unix = is_unix; s->do_nodelay = do_nodelay && !is_unix; chr->opaque = s; chr->chr_write = tcp_chr_write; chr->chr_close = tcp_chr_close; + chr->get_msgfd = tcp_get_msgfd; if (is_listen) { s->listen_fd = fd; diff --git a/qemu-char.h b/qemu-char.h index e1aa8db..77d4eda 100644 --- a/qemu-char.h +++ b/qemu-char.h @@ -51,6 +51,7 @@ struct CharDriverState { int (*chr_write)(struct CharDriverState *s, const uint8_t *buf, int len); void (*chr_update_read_handler)(struct CharDriverState *s); int (*chr_ioctl)(struct CharDriverState *s, int cmd, void *arg); + int (*get_msgfd)(struct CharDriverState *s); IOEventHandler *chr_event; IOCanRWHandler *chr_can_read; IOReadHandler *chr_read; @@ -81,6 +82,7 @@ void qemu_chr_reset(CharDriverState *s); void qemu_chr_initial_reset(void); int qemu_chr_can_read(CharDriverState *s); void qemu_chr_read(CharDriverState *s, uint8_t *buf, int len); +int qemu_chr_get_msgfd(CharDriverState *s); void qemu_chr_accept_input(CharDriverState *s); void qemu_chr_info(Monitor *mon); -- 1.6.2.5

1 0

[libvirt] libvirt-java, deprecated methods?
by Thomas Treutner 21 Jul '09

21 Jul '09

Hi, I'm playing around with libvirt-0.6.5 and libvirt-java-0.2.1 these days. It looks very interesting, but I think I may have found a bug: .... conn = new Connect("xen://node02", false); .... conn.setDom0Memory(512000) gives: exception caught:org.libvirt.LibvirtException: invalid domain pointer in virDomainSetMemory level:VIR_ERR_ERROR code:VIR_ERR_INVALID_DOMAIN domain:VIR_FROM_DOM hasConn:false hasDom:false hasNet:false message:invalid domain pointer in virDomainSetMemory str1:invalid domain pointer in %s str2:virDomainSetMemory str3:null int1:0 int2:0 Furthermore, f.e. conn.GetHypervisorVersion(conn.getType()) gives: exception caught:org.libvirt.LibvirtException: this function is not supported by the hypervisor: Xen level:VIR_ERR_ERROR code:VIR_ERR_NO_SUPPORT domain:VIR_FROM_NONE hasConn:false hasDom:false hasNet:false message:this function is not supported by the hypervisor: Xen str1:this function is not supported by the hypervisor: %s str2:Xen str3:null int1:0 int2:0 But Domain dom0=conn.domainLookupByID(0); //domainLookupByName("Domain-0"); too DomainInfo dom0info = dom0.getInfo(); .... dom0.setMemory(256000); .... works without glitches and does the job (btw, I can't see any effect of setMaxMemory() on Xen?) Is it possible that there are some inconsistences between libvirt-java and libvirt? Since libvirt-java wasn't updated approx. 1y, I assume that. Thanks for feedback, kr, thomas

3 5

[libvirt] [PATCH] Extra filesystem options during backend filesystem mount.
by Harshavardhana 21 Jul '09

21 Jul '09

New option index added to support -o options for various netfs. Currently added an option for glusterfs. --- src/storage_backend_fs.c | 32 ++++++++++++++++++++++++++++++++ 1 files changed, 32 insertions(+), 0 deletions(-) diff --git a/src/storage_backend_fs.c b/src/storage_backend_fs.c index c3d66b5..606fb47 100644 --- a/src/storage_backend_fs.c +++ b/src/storage_backend_fs.c @@ -604,6 +604,7 @@ static int virStorageBackendFileSystemMount(virConnectPtr conn, virStoragePoolObjPtr pool) { char *src; + char *options; const char **mntargv; /* 'mount -t auto' doesn't seem to auto determine nfs (or cifs), @@ -611,6 +612,10 @@ virStorageBackendFileSystemMount(virConnectPtr conn, * accommodate this */ int netauto = (pool->def->type == VIR_STORAGE_POOL_NETFS && pool->def->source.format == VIR_STORAGE_POOL_NETFS_AUTO); + int glusterfs = (pool->def->type == VIR_STORAGE_POOL_NETFS && + pool->def->source.format == VIR_STORAGE_POOL_NETFS_GLUSTERFS); + + int option_index; int source_index; const char *netfs_auto_argv[] = { @@ -632,9 +637,26 @@ virStorageBackendFileSystemMount(virConnectPtr conn, NULL, }; + const char *glusterfs_argv[] = { + MOUNT, + "-t", + pool->def->type == VIR_STORAGE_POOL_FS ? + virStoragePoolFormatFileSystemTypeToString(pool->def->source.format) : + virStoragePoolFormatFileSystemNetTypeToString(pool->def->source.format), + NULL, + "-o", + NULL, + pool->def->target.path, + NULL, + }; + if (netauto) { mntargv = netfs_auto_argv; source_index = 1; + } else if (glusterfs) { + mntargv = glusterfs_argv; + source_index = 3; + option_index = 5; } else { mntargv = fs_argv; source_index = 3; @@ -670,6 +692,12 @@ virStorageBackendFileSystemMount(virConnectPtr conn, } if (pool->def->type == VIR_STORAGE_POOL_NETFS) { + if (pool->def->source.format = VIR_STORAGE_POOL_NETFS_GLUSTERFS) { + if (virAsprintf(&options, "direct-io-mode=1") == -1) { + virReportOOMError(conn); + return -1; + } + } if (virAsprintf(&src, "%s:%s", pool->def->source.host.name, pool->def->source.dir) == -1) { @@ -685,6 +713,10 @@ virStorageBackendFileSystemMount(virConnectPtr conn, } mntargv[source_index] = src; + if (glusterfs) { + mntargv[option_index] = options; + } + if (virRun(conn, mntargv, NULL) < 0) { VIR_FREE(src); return -1; -- 1.6.0.6

3 6

[libvirt] Workaround to start xen inside a buggy kvm
by Thomas Fricke 21 Jul '09

21 Jul '09

Buggy kvm crashes with kernel panic if xen started inside a kvm machine. kvm needs the -no-kvm-pit, this argument is not available with libvirt. I have tested the following workaround on ubuntu 9.04 Rename kvm cd /usr/bin mv kvm kvm.orig Use the following script instead the original kvm #!/bin/sh exec kvm.orig -no-kvm-pit $* Do not forget chmod +x kvm Have fun, Thomas

1 0

[libvirt] [PATCH 0/6] virInterface bug fixes, test driver, and netcf backend driver
by Laine Stump 21 Jul '09

21 Jul '09

This series of patches first corrects a couple of bugs/deficiencies in existing virInterface code, then adds test and netcf-based backend interface drivers.

3 25

[libvirt] [patch] initialize a local variable in qemudOpenMonitorUnix()
by Jun Koi 21 Jul '09

21 Jul '09

This patch initializes a local variable in qemudOpenMonitorUnix(), thus also eliminates a compilation warning. diff --git a/src/qemu_driver.c b/src/qemu_driver.c index 00dc6e5..d2db1a2 100644 --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -874,7 +874,7 @@ qemudOpenMonitorUnix(virConnectPtr conn, struct sockaddr_un addr; int monfd; int timeout = 3; /* In seconds */ - int ret, i; + int ret, i = 0; if ((monfd = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) { virReportSystemError(conn, errno,

2 1

[libvirt] [PATCH] rpm spec cleanup and split off client only requirements
by Daniel Veillard 21 Jul '09

21 Jul '09

This early patch changes the spec file to: - remove the old Obsoletes: livir and such, that was 3 years ago - make sure the inter packages require %{version}-%{release} and not just %{version} - compresses the ChangeLog which is getting close to a megabyte - split out a separate libvirt-client binary rpm including only what is needed to access the libvirt service from a client viewpioint. The main package requires the client anyway So I moved in the client: - the shared library, I guess it's uncontroversial - the virsh/virt-xml-validate binaries, I guess it it doesn't block i386/x86_64 to be parallel installable that's fine - the man pages, obvious based on previous - the proxy when built, more controversial, it assumes that there is a service running locally so could be left on the main package - sasl and policykit data, more controversial, I guess it's better to associate them with the client - the schemas to allow checking on the client Some of this is still open, including the name of the subpackage which I initially expected to be libvirt-libs but in retrospect maybe -client will be more in line with the content, Comments welcome :-) Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel(a)veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/

2 2

[libvirt] [PATCH 0/7] Support cgroups in QEMU driver
by Daniel P. Berrange 20 Jul '09

20 Jul '09

This series of patches introduces cgroups support to the QEMU driver. At this time it uses the 'devices' controller to whitelist block devices for QEMU guests to prevent unauthorized access. It uses the 'cpu_shares' controller to allow schedular tunables on a per guest basis. In the future we should use the 'memory' controller to enforce the limit set in the balloon driver ie if the guest does not honour the balloon request, then force the guest into swap, avoiding host overcommit. There is also some refactoring of the cgroups code to remove the assumption that libvirtd is starting in the root cgroup, remove the requirement that all controllers be active, and allow for use in non-privileged drivers. Daniel P. Berrange (7): Use enums for cgroup controller types / labels Use virFileReadAll/virFileWriteStr for key cgroup read/write helpers Make cgroups a little more efficient Place every QEMU guest in a private cgroup Implement schedular tunables API using cgroups Use cgroups for block device whitelisting in QEMU guests Refactor cgroups to allow a group per driver to be managed directly src/cgroup.c | 861 ++++++++++++++++++++++++++-------------------- src/cgroup.h | 28 +- src/libvirt_private.syms | 1 + src/lxc_conf.h | 2 + src/lxc_controller.c | 19 +- src/lxc_driver.c | 25 +- src/qemu_conf.h | 2 + src/qemu_driver.c | 355 ++++++++++++++++++- src/util.c | 27 ++- src/util.h | 2 + 10 files changed, 904 insertions(+), 420 deletions(-)

2 17

[libvirt] changing files in /var/lib/libvirt/images
by Gene Czarcinski 20 Jul '09

20 Jul '09

Assume I am working with virtual disk image files in /var/lib/libvirt/images such as removing a file, defining a new file (with qemi-img create or qemu-img convert), or simply copying a file ("cp") from elsewhere such as VMware. When I do this, Guest definition or adding hardware to an existing guest under virt- manager does not "see" the file change. That is, until I terminate virt- manager and restart /etc/init.d/libvirtd. Is there some better/easier/less-disruptive means of notifying "whatever" that the list of files has changed? I noticed that /etc/init.d/libvirtd force-reload has no effect. I also notice that restarting libvirtd and/or terminating and restarting virt- manager while there are guests running really screws things up. If such a "means" does not exist, should one be created? [should this be an RFE in bugzilla?] Gene

2 2

[libvirt] Create a new VM from an existent image?
by Jun Koi 19 Jul '09

19 Jul '09

Hi, To use libvirt, I am strying to use virt-install to create a new VM from an existent (KVM) VM image. I did the following steps: 1) run libvirtd with "libvirtd -d" (I use the latest libvirt git tree) 2) Run virt-install (0.400.3) with my existent image in img/img.winxp # virt-install --import -n winxp -f img/img.xp3 --vnc --hvm --accelerate --ram 800 However, this step returns error like: "ERROR Could not find usable default libvirt connection." Actually I am not sure that is the right way to use the new --import option with virt-install. So that might be the problem? All the code are updated, and compiled from source on Ubuntu 8.04. Any help? Thanks a lot, J

2 4

[libvirt] [PATCH] qemu: Try multiple times to open unix monitor socket
by Cole Robinson 17 Jul '09

17 Jul '09

Unlike the pty monitor (which we know exists since we scrape its path from stdout), we have no way of knowing that the unix monitor socket should exist/ be initialized. As a result, some of my KVM guests randomly fail to start on F10 host. Try to open the unix socket in a 3 second timeout loop. Ignore EACCES (path does not exist if a first time run) and ECONNREFUSED (leftover socket from a previous run hasn't been removed yet). Fixes things for me. --- src/qemu_driver.c | 22 +++++++++++++++++++++- 1 files changed, 21 insertions(+), 1 deletions(-) diff --git a/src/qemu_driver.c b/src/qemu_driver.c index e2b7acb..200718b 100644 --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -874,6 +874,8 @@ qemudOpenMonitorUnix(virConnectPtr conn, { struct sockaddr_un addr; int monfd; + int timeout = 3; /* In seconds */ + int ret, i; if ((monfd = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) { virReportSystemError(conn, errno, @@ -885,10 +887,28 @@ qemudOpenMonitorUnix(virConnectPtr conn, addr.sun_family = AF_UNIX; strncpy(addr.sun_path, monitor, sizeof(addr.sun_path)); - if (connect(monfd, (struct sockaddr *) &addr, sizeof(addr)) < 0) { + do { + ret = connect(monfd, (struct sockaddr *) &addr, sizeof(addr)); + + if (ret == 0) + break; + + if (errno == EACCES || errno == ECONNREFUSED) { + /* EACCES : Socket may not have shown up yet + * ECONNREFUSED : Leftover socket hasn't been removed yet */ + continue; + } + virReportSystemError(conn, errno, "%s", _("failed to connect to monitor socket")); goto error; + + } while ((++i <= timeout*5) && (usleep(.2 * 1000000) <= 0)); + + if (ret != 0) { + virReportSystemError(conn, errno, "%s", + _("monitor socket did not show up.")); + goto error; } if (qemudOpenMonitorCommon(conn, driver, vm, monfd, reconnect) < 0) -- 1.6.0.6

4 5

[libvirt] adding scsi disk failed (code=9)
by Massimo Canonico 17 Jul '09

17 Jul '09

Hi, I'm newbie and sorry if this question was already post in this ML (is there any search engine for the ML archive?). I'm trying to attach a volume in my instance. The instance is running and the volume is available, but i got this error: ---- libvirt: operation failed: adding scsi disk failed (code=9) virDomainAttachDevice() failed (err=-1) XML=<disk type='block'><driver name='phy'/><source dev='/dev/etherd/e0.8'/><target dev='sdb'/></disk> doAttachVolume() failed error=1 ---- I'm not sure where I can start to debug this error. I'm using Fedora 11, x86_64, with these packages [root@minicloud eucalyptus]# rpm -qa | grep libv libvirt-devel-0.6.2-12.fc11.x86_64 libvorbis-1.2.0-7.fc11.x86_64 libvirt-0.6.2-12.fc11.x86_64 libvirt-python-0.6.2-12.fc11.x86_64 libvolume_id-141-3.fc11.x86_64 May you help me, please? Thanks in advance, M

4 5

[libvirt] [libvirt 0.6.4] Unable to set /domain/devices/serial[@port=1] as console?
by Charles Duffy 17 Jul '09

17 Jul '09

When I do a "virsh define" with the following: <serial type='file'> <source path='/path/to/serial.log'/> <target port='0'/> </serial> <serial type='pty'> <target port='1'/> </serial> <console type='pty'> <target port='1'/> </console> The result is this: <serial type='file'> <source path='/path/to/serial.log'/> <target port='0'/> </serial> <serial type='pty'> <target port='1'/> </serial> <console type='file'> <source path='/path/to/serial.log'/> <target port='0'/> </console> ...a markedly different configuration. Is this the result of a bug, or a known behavior I've failed to catch in the docs? Thanks!

2 1

[libvirt] Tentative schedule for 0.7.0
by Daniel Veillard 17 Jul '09

17 Jul '09

Basically I would like to freeze development at the end of next week and push at the end of the month. Since I plan to be away on 31, I will probably make the release on the 30th, but I think we can still apply new features until Friday next week. To try to help w.r.t. Fedora 12 schedule I will make a 0.7.0 beta release probably on the week-end of the 25 and push it in rawhide for parallel testing, Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel(a)veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/

1 0

[libvirt] One line logging fix
by Dave Allan 17 Jul '09

17 Jul '09

I noticed that I was incorrectly logging a debug message as an error in a success path in the node device code. Here's the fix. >From 0c65b2d1d90bf0d469ffe6d8a20a5566b359a61e Mon Sep 17 00:00:00 2001 From: David Allan <dallan(a)redhat.com> Date: Fri, 17 Jul 2009 09:57:40 -0400 Subject: [PATCH 1/1] Remove spurious log message A success path in part of the node device code contained a log message with severity ERROR; this fix makes it DEBUG. --- src/node_device_hal_linux.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/node_device_hal_linux.c b/src/node_device_hal_linux.c index ddb1eef..a119404 100644 --- a/src/node_device_hal_linux.c +++ b/src/node_device_hal_linux.c @@ -50,7 +50,7 @@ static int open_wwn_file(const char *prefix, /* fd will be closed by caller */ if ((*fd = open(wwn_path, O_RDONLY)) != -1) { - VIR_ERROR(_("Opened WWN path '%s' for reading"), + VIR_DEBUG(_("Opened WWN path '%s' for reading"), wwn_path); } else { VIR_ERROR(_("Failed to open WWN path '%s' for reading"), -- 1.6.0.6

2 1

[libvirt] Guest shutdown script
by Ján ONDREJ (SAL) 17 Jul '09

17 Jul '09

Hello, when host server is shutting down, it's goot to shutdown also guests, not only kill them. Host shutdown can be touched by an admin, but also can be started by an UPS daemon or similar action. Then kill of guests is not a best solution. I have my own script, which can shutdown guest automatically: http://www.salstar.sk/pub/salpack/etc/rc.d/init.d/virt-shutdown You need to start this rc script after installation, otherwise it will not do anything. Another init script with save/restore ability is here: http://marc.info/?l=kvm&m=124775655529836&w=4 I'm not using this so I don't know, if this will work well on Fedora. A good solution can also be, if libvirt init script can shutdown guests autoamtically. It starts them, when system is booting, but does not stop them. I know, that libvirt sometimes need to be restarted without touching guests, but you also can avoid this, if you shutdown guest only if one of: - current runlevel is 0 (halt) or 6 (reboot). You can get current runlevel by running: runlevel | cut -d" " -f2 - if "service libvirt shutdown" is called, which is not very systematic, but can work well Just please do not create a new package only for guest shutdowns. :-) I am not a member of this list, so please cc me. SAL

1 0

[libvirt] How to pass not xml parameters from Libvirt to Xen in xml definition
by Alexandre CABROL PERALES 17 Jul '09

17 Jul '09

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi everybody, i need to define cpuid parameter for a specific Xen Virtual Machine. This parameters doesn't exist in libvirt definition. So how can i add this "unknow" parameter in virtual machine's xml definition to be able to manage it from virt-manager? regards, - -- Alexandre CABROL PERALES - -- Ingenieur Securite des Systemes d'Information Mob. : 06.98.82.03.06 Mail : alexandre.cabrol(a)artal.fr Key fingerprint = 1E6B B8DF 5001 A6A8 E057 9D31 7B3B EAB1 4AE4 8953 - -- ARTAL Technologies Rue Pierre-Gilles de Gennes Ens."La Rue", Bat. 9, BP 38138 31681 Labege cedex Tel. : 05.61.00.39.30 Fax : 05.61.00.20.43 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkpgSWsACgkQezvqsUrkiVMqiwCgkigm/WZ5TPToQ7RzdY0R6UJz tEAAoKIcdZ2fO5PdqZdcrzuTj6Ev4yh4 =8rlJ -----END PGP SIGNATURE-----

2 2