June 2015 - Devel - libvirt List Archives

[libvirt] poor virtio-scsi performance
by Vasiliy Tolstov 08 Jun '15

08 Jun '15

Hi all! I suspected poor performance of virtio-scsi driver. I did a few tests: Host machine: linux 3.19.1, QEMU emulator version 2.3.0 Guest machine: linux 4.0.4 part of domain xml: <emulator>/usr/bin/kvm</emulator> <disk type='block' device='disk'> <driver name='qemu' type='raw' cache='none' io='native' discard='unmap'/> <source dev='/dev/ram0'/> <backingStore/> <target dev='sda' bus='scsi'/> <alias name='scsi0-0-0-1'/> <address type='drive' controller='0' bus='0' target='0' unit='1'/> </disk> /dev/ram0 I got by running `modprobe brd rd_size=$((5*1024*1024))` on host machine. fio conf: [readtest] blocksize=4k filename=/dev/sdb (/dev/ram0 whe test from host machine) rw=randread direct=1 buffered=0 ioengine=libaio iodepth=32 results: from host: bw=1594.6MB/s, iops=408196, clat=76usec from guest: bw=398MB/s, iops=99720, clat=316usec Both host and guest system I boot with `scsi_mod.use_blk_mq=Y`. Why difference in 4 times?! -- Vasiliy Tolstov, e-mail: v.tolstov(a)selfip.ru

2 1

[libvirt] [PATCH v2] utiltest: Use int8_t instead of char.
by Michal Privoznik 08 Jun '15

08 Jun '15

Not every architecture out there has 'char' signed by default. For instance, my arm box has it unsigned by default: $ gcc -dM -E - < /dev/null | grep __CHAR_UNSIGNED__ #define __CHAR_UNSIGNED__ 1 Therefore, after 65c61e50 the test if failing for me. Problem is, we are trying to assign couple of negative values into char assuming some will overflow and some don't. That can't be the case if 'char' is unsigned by default. Lets use more explicit types instead: int8_t and uint8_t where is no ambiguity. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- tests/utiltest.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/utiltest.c b/tests/utiltest.c index 3a1f8eb..98c689d 100644 --- a/tests/utiltest.c +++ b/tests/utiltest.c @@ -184,8 +184,8 @@ static int testOverflowCheckMacro(const void *data ATTRIBUTE_UNUSED) { long long tmp; - unsigned char luchar; - char lchar; + uint8_t luchar; + int8_t lchar; TEST_OVERFLOW(luchar, 254, false); TEST_OVERFLOW(luchar, 255, false); -- 2.3.6

2 1

[libvirt] [PATCH] apibuild: Fix indentation
by Jiri Denemark 08 Jun '15

08 Jun '15

Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com> --- docs/apibuild.py | 48 ++++++++++++++++++++++++------------------------ 1 file changed, 24 insertions(+), 24 deletions(-) diff --git a/docs/apibuild.py b/docs/apibuild.py index 9fa9361..18278db 100755 --- a/docs/apibuild.py +++ b/docs/apibuild.py @@ -1356,32 +1356,32 @@ class CParser: token = self.token() return token elif token[0] == "name": - self.cleanupComment() - if name is not None: - if self.comment is not None: - comment = string.strip(self.comment) - self.comment = None - self.enums.append((name, value, comment)) - name = token[1] - comment = "" + self.cleanupComment() + if name is not None: + if self.comment is not None: + comment = string.strip(self.comment) + self.comment = None + self.enums.append((name, value, comment)) + name = token[1] + comment = "" + token = self.token() + if token[0] == "op" and token[1][0] == "=": + value = "" + if len(token[1]) > 1: + value = token[1][1:] token = self.token() - if token[0] == "op" and token[1][0] == "=": - value = "" - if len(token[1]) > 1: - value = token[1][1:] - token = self.token() - while token[0] != "sep" or (token[1] != ',' and - token[1] != '}'): - value = value + token[1] - token = self.token() - else: - try: - value = "%d" % (int(value) + 1) - except: - self.warning("Failed to compute value of enum %s" % (name)) - value="" - if token[0] == "sep" and token[1] == ",": + while token[0] != "sep" or (token[1] != ',' and + token[1] != '}'): + value = value + token[1] token = self.token() + else: + try: + value = "%d" % (int(value) + 1) + except: + self.warning("Failed to compute value of enum %s" % (name)) + value="" + if token[0] == "sep" and token[1] == ",": + token = self.token() else: token = self.token() return token -- 2.4.2

2 1

[libvirt] [PATCH 0/2] Couple of unrelated fixes
by Michal Privoznik 08 Jun '15

08 Jun '15

*** BLURB HERE *** Michal Privoznik (2): utils: Return proper value for virGet{User,Group}ID utiltest: Don't assume 'char' is always signed src/util/virutil.c | 4 ++-- tests/utiltest.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) -- 2.3.6

3 5

[libvirt] the error is bug？
by cloudbuglist 08 Jun '15

08 Jun '15

Hi dear! Today I came across a problem, the following is a log and software versions. soft version: os:centos 6.5 libvirt:0.10.29 qemu:0.12.415 2015-05-28 03:28:58.001+0000: 20919: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:30:37.000+0000: 20918: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:30:37.000+0000: 20918: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:31:01.000+0000: 20919: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:31:01.000+0000: 20919: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:32:40.000+0000: 20921: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:32:40.000+0000: 20921: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:33:06.000+0000: 20918: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:33:06.000+0000: 20918: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:33:42.009+0000: 20916: error : virNetSocketReadWire:1194 : End of file while reading data: Input/output error 2015-05-28 03:34:02.000+0000: 20921: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:34:02.000+0000: 20921: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:34:19.001+0000: 20920: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:34:19.001+0000: 20920: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:34:43.384+0000: 20916: error : virNetSocketReadWire:1194 : End of file while reading data: Input/output error 2015-05-28 03:34:47.000+0000: 20918: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:34:47.000+0000: 20918: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:34:52.000+0000: 20920: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:34:52.000+0000: 20920: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock 2015-05-28 03:35:09.001+0000: 20921: warning : qemuDomainObjBeginJobInternal:839 : Cannot start job (query, none) for domain i-2-437-VM; current job is (query, none) owned by (20917, 0) 2015-05-28 03:35:09.001+0000: 20921: error : qemuDomainObjBeginJobInternal:843 : Timed out during operation: cannot acquire state change lock

2 1

[libvirt] [PATCH] util: Properly return error from virGetUserID and virGetGroupID stubs
by Peter Krempa 08 Jun '15

08 Jun '15

The stubs for the two functions that are compiled on platforms that don't have HAVE_GETPWUID_R and friends defined do not return error but report an error message. The calling code then assumes that the @uid or @gid arguments were filled, which is not the case in the stubs. --- This should fix the mingw build that complains that @theuid in one of the callers is not initialized. src/util/virutil.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/util/virutil.c b/src/util/virutil.c index e479cce..cddc78a 100644 --- a/src/util/virutil.c +++ b/src/util/virutil.c @@ -1316,7 +1316,7 @@ int virGetUserID(const char *name ATTRIBUTE_UNUSED, virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("virGetUserID is not available")); - return 0; + return -1; } @@ -1326,7 +1326,7 @@ int virGetGroupID(const char *name ATTRIBUTE_UNUSED, virReportError(VIR_ERR_INTERNAL_ERROR, "%s", _("virGetGroupID is not available")); - return 0; + return -1; } int -- 2.4.1

2 1

[libvirt] [PATCH] util: process: fix build on FreeBSD
by Roman Bogorodskiy 08 Jun '15

08 Jun '15

Commit 825df8c3 refactored virProcess{Set,Get}Affinity routines, however broke BSD implementation because of the incorrect variable name. Fix build by using a proper variable name. Pushing as trivial and build break fix. --- src/util/virprocess.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/util/virprocess.c b/src/util/virprocess.c index a38cb75..59598ed 100644 --- a/src/util/virprocess.c +++ b/src/util/virprocess.c @@ -566,7 +566,7 @@ virProcessGetAffinity(pid_t pid) return NULL; } - if (!(*map = virBitmapNew(sizeof(mask) * 8))) + if (!(ret = virBitmapNew(sizeof(mask) * 8))) return NULL; for (i = 0; i < sizeof(mask) * 8; i++) -- 2.4.2

1 0

[libvirt] bulk stats for block and cpu
by Vasiliy Tolstov 08 Jun '15

08 Jun '15

Hi. Is i remember sometimes ago someone implements bulk stats for libvirt (cpu, block, network..?) to able to get all needed info with one api call. But i can't find api in libvirt docs, does anybody helps me and send link to docs, ii want to get info about block stats for 100-200 domains. -- Vasiliy Tolstov, e-mail: v.tolstov(a)selfip.ru

2 2

[libvirt] ANNOUNCE: virt-manager 1.2.1 released
by Cole Robinson 06 Jun '15

06 Jun '15

I'm happy to announce the release of virt-manager 1.2.1! virt-manager is a desktop application for managing KVM, Xen, and LXC virtualization via libvirt. The release can be downloaded from: http://virt-manager.org/download/ The direct download links are: http://virt-manager.org/download/sources/virt-manager/virt-manager-1.2.1.ta… This release is mostly for bug scooping up some bugfixes. Including: - Fix connecting to older libvirt versions (Michał Kępień) - Fix connecting to VM console with non-IP hostname (Giuseppe Scrivano) - Fix addhardware/create wizard errors when a nodedev disappears - Fix adding a second cdrom via customize dialog Thanks to everyone who has contributed to this release through testing, bug reporting, submitting patches, and otherwise sending in feedback! Thanks, Cole

1 0

[libvirt] [RFC] driver's persistent connection closed event
by Nikolay Shirokovskiy 06 Jun '15

06 Jun '15

Hello. I have a question concerning hypervisor driver "persistent connection closed" event mechanics. Currently the only driver that have such a connection (or at least notify of such event) is the remote driver. Accordingly we don't have interface for this event at driver level. Parallels driver actually have a persistent connection. Imagine we have a libvirt connection to parallels driver through remote driver, if parallels internal connection is closed we can catch it in daemon but the only option to notify remote driver of this event would be closing of remote driver to daemon connection which will in turn trigger connection closed event deliver to client. Thus event will be delivered but at cost of forcedly closed connection to daemon. I suspect the described solution have architectural flaws so i suggest move interface of "persistent connection closed" event to driver level which will give us a chance to deliver this event without side effects. We don't need to implement this interface for most drivers and clients could handle this well as in this case "unimplemented" would be a synonym of "have not persistent connection". Alternatively we could silently do nothing if driver does not support this interface so clients would not break.

1 0