- CIM - libvirt List Archives

virNodeGetFreeMemory() returns bytes with Xen driver
by Tyrel Datwyler 17 Apr '09

17 Apr '09

The libvirt API states that the return value of virNodeGetFreeMemory() is in kilobytes. When making this call against a Xen hypervisor and comparing the returned value with what is listed by "xm info" it appears that libvirt is returning the free memory in bytes and not kilobytes. Can anybody confirm this behavior? This code snippet: conn = virConnectOpen("xen:///localhost/"); hostmem = virNodeGetFreeMemory(conn); fprintf(stdout, "Host free memory = %llu\n", hostmem); produces the following output: Host free memory = 3236012032 The following comes from running "xm info" on the Xen host: free_memory : 3086 The libvirt value is convertable into the "xm" returned value if bytes is assumed. Regards, Tyrel Datwyler Linux Software Engineer eServer Systems Management IBM Systems and Technology Group Email: tyreld(a)us.ibm.com External: (503) 578-3489, T/L: 775-3489

3 2

[PATCH] [TEST] Fix SettingsDefineCapabilities/01_forward.py with the provider's changes of the output of ac association when using DiskPool as input
by yunguol＠cn.ibm.com 16 Apr '09

16 Apr '09

# HG changeset patch # User Guolian Yun <yunguol(a)cn.ibm.com> # Date 1239851971 25200 # Node ID de971738caf9e9b3eae3c3fe8dc65cc7114a7ff1 # Parent 4ec367c94c356de7fac5a19ffe215c316d0cdcd1 [TEST] Fix SettingsDefineCapabilities/01_forward.py with the provider's changes of the output of ac association when using DiskPool as input Tested for KVM with current sources and rpm Signed-off-by: Guolian Yun<yunguol(a)cn.ibm.com diff -r 4ec367c94c35 -r de971738caf9 suites/libvirt-cim/cimtest/SettingsDefineCapabilities/01_forward.py --- a/suites/libvirt-cim/cimtest/SettingsDefineCapabilities/01_forward.py Wed Apr 08 02:22:53 2009 -0700 +++ b/suites/libvirt-cim/cimtest/SettingsDefineCapabilities/01_forward.py Wed Apr 15 20:19:31 2009 -0700 @@ -64,10 +64,12 @@ from XenKvmLib.classes import get_typed_class from XenKvmLib.common_util import print_field_error from XenKvmLib.const import get_provider_version +from XenKvmLib.pool import enum_volumes platform_sup = ['Xen', 'KVM', 'XenFV', 'LXC'] libvirt_rasd_template_changes = 707 libvirt_rasd_new_changes = 805 +libvirt_rasd_dpool_changes = 839 memid = "MemoryPool/0" procid = "ProcessorPool/0" @@ -184,8 +186,12 @@ if curr_cim_rev >= libvirt_rasd_new_changes: exp_len = 16 if virt == 'KVM': - if curr_cim_rev >= libvirt_rasd_new_changes: + if curr_cim_rev >= libvirt_rasd_new_changes and \ + curr_cim_rev < libvirt_rasd_dpool_changes: exp_len = 8 + if curr_cim_rev >= libvirt_rasd_dpool_changes: + volumes = enum_volumes(virt, server) + exp_len = volumes * 4 if len(assoc_info) != exp_len: logger.error("%s returned %i ResourcePool objects instead"

1 0

[PATCH] [TEST] #2 Fix 03_hs_to_settdefcap.py with the provider's changes of the output of ac association when using DiskPool as input
by yunguol＠cn.ibm.com 16 Apr '09

16 Apr '09

# HG changeset patch # User Guolian Yun <yunguol(a)cn.ibm.com> # Date 1239851524 25200 # Node ID 352e8509493cfce2620bca94798a1012c63cb410 # Parent 4ec367c94c356de7fac5a19ffe215c316d0cdcd1 [TEST] #2 Fix 03_hs_to_settdefcap.py with the provider's changes of the output of ac association when using DiskPool as input Updates from 1 to 2: Query libvirt to get the number of DiskRASD instances instead of hardcode Tested for KVM with current sources and rpm Signed-off-by: Guolian Yun<yunguol(a)cn.ibm.com> diff -r 4ec367c94c35 -r 352e8509493c suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py --- a/suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py Wed Apr 08 02:22:53 2009 -0700 +++ b/suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py Wed Apr 15 20:12:04 2009 -0700 @@ -51,12 +51,14 @@ from XenKvmLib.test_xml import testxml from XenKvmLib.test_doms import destroy_and_undefine_all from XenKvmLib.const import get_provider_version +from XenKvmLib.pool import enum_volumes sup_types = ['Xen', 'KVM', 'XenFV', 'LXC'] test_dom = "domgst_test" test_vcpus = 1 libvirt_rasd_template_changes = 707 libvirt_rasd_new_changes = 805 +libvirt_rasd_dpool_changes = 839 def setup_env(server, virt="Xen"): status = PASS @@ -227,8 +229,12 @@ if curr_cim_rev >= libvirt_rasd_new_changes: exp_len = 16 if virt == 'KVM': - if curr_cim_rev >= libvirt_rasd_new_changes: + if curr_cim_rev >= libvirt_rasd_new_changes and \ + curr_cim_rev < libvirt_rasd_dpool_changes: exp_len = 8 + if curr_cim_rev >= libvirt_rasd_dpool_changes: + volumes = enum_volumes(virt, server) + exp_len = volumes * 4 if len(assoc_info) != exp_len: logger.error("'%s' returned %i RASD objects instead of %i", diff -r 4ec367c94c35 -r 352e8509493c suites/libvirt-cim/lib/XenKvmLib/pool.py --- a/suites/libvirt-cim/lib/XenKvmLib/pool.py Wed Apr 08 02:22:53 2009 -0700 +++ b/suites/libvirt-cim/lib/XenKvmLib/pool.py Wed Apr 15 20:12:04 2009 -0700 @@ -26,6 +26,8 @@ from XenKvmLib.classes import get_typed_class from XenKvmLib.const import get_provider_version from XenKvmLib.enumclass import EnumInstances +from VirtLib.utils import run_remote +from XenKvmLib.xm_virt_util import virt2uri input_graphics_pool_rev = 757 @@ -78,3 +80,16 @@ return pool_insts, PASS +def enum_volumes(virt, server): + volume = 0 + cmd = "virsh -c %s vol-list %s | sed -e '1,2 d' -e '$ d'" % \ + (virt2uri(virt), 'cimtest-diskpool') + ret, out = run_remote(server ,cmd) + if ret != 0: + return None + lines = out.split("\n") + for line in lines: + volume = volume + 1 + + return volume +

1 0

[PATCH] [TEST]#2 Fix 03_hs_to_settdefcap.py with the provider's changes of the output of ac association when using DiskPool as input
by yunguol＠cn.ibm.com 16 Apr '09

16 Apr '09

# HG changeset patch # User Guolian Yun <yunguol(a)cn.ibm.com> # Date 1239849166 25200 # Node ID 7b81278fbdd9d4337ac7422e98bcfebc93b70e64 # Parent 4ec367c94c356de7fac5a19ffe215c316d0cdcd1 [TEST]#2 Fix 03_hs_to_settdefcap.py with the provider's changes of the output of ac association when using DiskPool as input Updates from 1 to 2: Query libvirt to get the number of DiskRASD instances and then compare to the result Tested for KVM with current sources and rpm Signed-off-by: Guolian Yun<yunguol(a)cn.ibm.com> diff -r 4ec367c94c35 -r 7b81278fbdd9 suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py --- a/suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py Wed Apr 08 02:22:53 2009 -0700 +++ b/suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py Wed Apr 15 19:32:46 2009 -0700 @@ -40,7 +40,9 @@ # Feb 13 2008 import sys -from VirtLib.live import full_hostname +from VirtLib.live import full_hostname +from VirtLib.utils import run_remote +from XenKvmLib.xm_virt_util import virt2uri from XenKvmLib.common_util import get_host_info from XenKvmLib.assoc import Associators from XenKvmLib.vxml import XenXML, KVMXML, get_class @@ -57,6 +59,7 @@ test_vcpus = 1 libvirt_rasd_template_changes = 707 libvirt_rasd_new_changes = 805 +libvirt_rasd_dpool_changes = 839 def setup_env(server, virt="Xen"): status = PASS @@ -216,6 +219,16 @@ curr_cim_rev, changeset = get_provider_version(virt, server) exp_len = 4 + + volume = 0 + cmd = "virsh -c %s vol-list %s" % (virt2uri(server), 'cimtest-diskpool') + ret, out = run_remote(server ,cmd) + if ret != 0: + return None + lines = out.split("\n") + for line in lines: + volume = volume + 1 + if 'DiskPool' in ap['InstanceID']: # For Diskpool, we have info 1 for each of Min, Max, # default, Increment and 1 for each of PV and FV @@ -227,8 +240,11 @@ if curr_cim_rev >= libvirt_rasd_new_changes: exp_len = 16 if virt == 'KVM': - if curr_cim_rev >= libvirt_rasd_new_changes: + if curr_cim_rev >= libvirt_rasd_new_changes and \ + curr_cim_rev < libvirt_rasd_dpool_changes: exp_len = 8 + if curr_cim_rev >= libvirt_rasd_dpool_changes: + exp_len = volume if len(assoc_info) != exp_len: logger.error("'%s' returned %i RASD objects instead of %i",

2 1

Release of libvirt-cim 0.5.5
by Kaitlin Rupert 16 Apr '09

16 Apr '09

A new release of libvirt-cim is now available at: ftp://libvirt.org/libvirt-cim/ * Features: - Basic network pool creation and deletion - Basic storage pool creation and deletion (directory pools only) - Storage volumes in a pool represented as template ResourceAllocationSettingData - Advertise profiles via SLP * Bug fixes: - Improved reporting: libvirt error is now included in error message returned to the user - ElementConformsToProfile association now supports the RegisteredProfile <--> AllocationCapabilities - Prevent seg fault in get_typed_class() by returning is a NULL value is specified for refcn - Fix xmlgen so XML isn't generated for devices with an UNKNOWN device type - Set the device type of a resource to UNKNOWN after the call to detach_device(), otherwise detach_device() doesn't know what device type to handle - Allow user to specify guest's emulator and expose emulator via the guest's VirtualSystemSettingData - Improve randomization of MAC generation - Determine whether the host system supports KVM or QEMU guests. If KVM support is available, create a KVM guest. Otherwise, create a QEMU guest. - Fix some connection leaks and some libvirt pointer leaks - Be sure to specify the root/virt namespace (instead of root/interop) in ElementConformsToProfile when building the instance of HostSystem

1 0

[PATCH] Pass OP with root/virt NS to get_host() in ECTP
by Kaitlin Rupert 15 Apr '09

15 Apr '09

# HG changeset patch # User Kaitlin Rupert <karupert(a)us.ibm.com> # Date 1239679403 25200 # Node ID c9160bb4698b629c7832efcd32f6641fd74449de # Parent 8d0ce3b79aaacde9a7b25d8e2fe2eb79878229ee Pass OP with root/virt NS to get_host() in ECTP When calling get_host() in ECTP, be sure to pass object path with root/virt namespace instead of root/interop. get_host() uses the namespace of the ref when creating the instance, so this needs to be set properly. Signed-off-by: Kaitlin Rupert <karupert(a)us.ibm.com> diff -r 8d0ce3b79aaa -r c9160bb4698b src/Virt_ElementConformsToProfile.c --- a/src/Virt_ElementConformsToProfile.c Wed Apr 08 22:22:42 2009 -0400 +++ b/src/Virt_ElementConformsToProfile.c Mon Apr 13 20:23:23 2009 -0700 @@ -64,15 +64,7 @@ if (class == NULL) return s; - if (STREQC(class, "HostSystem")) { - s = get_host(_BROKER, info->context, ref, &inst, false); - if (s.rc == CMPI_RC_OK) - inst_list_add(list, inst); - goto out; - } - - classname = get_typed_class(pfx_from_conn(conn), - class); + classname = get_typed_class(pfx_from_conn(conn), class); if (classname == NULL) { cu_statusf(_BROKER, &s, CMPI_RC_ERR_FAILED, @@ -83,7 +75,14 @@ op = CMNewObjectPath(_BROKER, CIM_VIRT_NS, classname, &s); if ((s.rc != CMPI_RC_OK) || CMIsNullObject(op)) goto out; - + + if (STREQC(class, "HostSystem")) { + s = get_host(_BROKER, info->context, op, &inst, false); + if (s.rc == CMPI_RC_OK) + inst_list_add(list, inst); + goto out; + } + en = CBEnumInstances(_BROKER, info->context , op, info->properties, &s); if (en == NULL) { cu_statusf(_BROKER, &s,

2 1

[PATCH] Add logic to delete storage pools
by Kaitlin Rupert 14 Apr '09

14 Apr '09

# HG changeset patch # User Kaitlin Rupert <karupert(a)us.ibm.com> # Date 1239296847 25200 # Node ID cdf4fa212c15f1b22b0b4b1d1535e1716e1e52d2 # Parent 3b7ba02cd34a4b91903e4d412fc7d5dd75779598 Add logic to delete storage pools. Older versions of libvirt don't support storage pool creation / deletion. So the pool deletion logic needs to return a different error message in the case of older versions. Signed-off-by: Kaitlin Rupert <karupert(a)us.ibm.com> diff -r 3b7ba02cd34a -r cdf4fa212c15 libxkutil/pool_parsing.c --- a/libxkutil/pool_parsing.c Tue Apr 07 21:54:43 2009 -0300 +++ b/libxkutil/pool_parsing.c Thu Apr 09 10:07:27 2009 -0700 @@ -138,9 +138,32 @@ err1: virNetworkFree(ptr); + + } else if (res_type == CIM_RES_TYPE_DISK) { +#if VIR_USE_LIBVIRT_STORAGE + virStoragePoolPtr ptr = virStoragePoolLookupByName(conn, name); + if (ptr == NULL) { + CU_DEBUG("Storage pool %s is not defined", name); + return 0; + } + + if (virStoragePoolDestroy(ptr) != 0) { + CU_DEBUG("Unable to destroy storage pool"); + goto err2; + } + + if (virStoragePoolUndefine(ptr) != 0) { + CU_DEBUG("Unable to undefine storage pool"); + goto err2; + } + + ret = 1; + + err2: + virStoragePoolFree(ptr); +#endif } - return ret; } diff -r 3b7ba02cd34a -r cdf4fa212c15 src/Virt_ResourcePoolConfigurationService.c --- a/src/Virt_ResourcePoolConfigurationService.c Tue Apr 07 21:54:43 2009 -0300 +++ b/src/Virt_ResourcePoolConfigurationService.c Thu Apr 09 10:07:27 2009 -0700 @@ -139,6 +139,18 @@ return msg; } + +static const char *_delete_pool(virConnectPtr conn, + const char *pool_name, + uint16_t type) +{ + const char *msg = NULL; + + if (destroy_pool(conn, pool_name, type) == 0) + msg = "Unable to destroy resource pool"; + + return msg; +} #else static const char *disk_rasd_to_pool(CMPIInstance *inst, struct virt_pool *pool, @@ -146,6 +158,13 @@ { return "Storage pool creation not supported in this version of libvirt"; } + +static const char *_delete_pool(virConnectPtr conn, + const char *pool_name, + uint16_t type) +{ + return "Storage pool deletion not supported in this version of libvirt"; +} #endif static const char *rasd_to_vpool(CMPIInstance *inst, @@ -379,6 +398,7 @@ CMPIObjectPath *pool = NULL; virConnectPtr conn = NULL; const char *poolid = NULL; + const char *msg = NULL; char *pool_name = NULL; uint16_t type; @@ -432,10 +452,12 @@ goto out; } - if (destroy_pool(conn, pool_name, type) == 0) { + msg = _delete_pool(conn, pool_name, type); + if (msg != NULL) { cu_statusf(_BROKER, &s, CMPI_RC_ERR_FAILED, - "Unable to destroy resource pool"); + "Storage pool deletion error: %s", msg); + goto out; }

2 1

[PATCH 0 of 3] Clean up leaks, ensure libvirt pointers are properly freed
by Kaitlin Rupert 14 Apr '09

14 Apr '09

These fixes resolve an occasional hang and and recv error messages seen when running with upstream libvirt. The recv error message is the same one seen in the test run from March 25th on KVM on Fedora rawhide: "libvir: Remote error : cannot recv data: Connection reset by peer"

2 4

Re: [Libvirt-cim] [PATCH] [TEST] #2 Add branch of err code to RPCS/04
by Kaitlin Rupert 14 Apr '09

14 Apr '09

> > supported. So, for libvirt-cim >= 837, this test should attempt to > > create network pool and verify that network pool is created. > > There are three arguments in CreateChildResourcePool: > ElementName, Settings and ParentPool. > What's the ParentPool on cimtest? Do I have to create a primordial > pool at first? Right now, the providers ignore the value specified for the ParentPool. Support for ParentPool still needs to be added. > > One more thing, the provider only support create networkpool now, > right? > > Creating network pools and directory based storage pools is supported.

2 1

Re: [Libvirt-cim] [PATCH] [TEST] Fix 03_hs_to_settdefcap.py with the provider's changes of the output of ac association when using DiskPool as input
by Kaitlin Rupert 13 Apr '09

13 Apr '09

> diff -r 6027a48279e6 -r 9ed17bb61735 > suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py > --- a/suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py Tue > Apr 07 06:29:37 2009 -0700 > +++ b/suites/libvirt-cim/cimtest/HostSystem/03_hs_to_settdefcap.py Sun > Apr 12 20:18:16 2009 -0700 > @@ -57,6 +57,7 @@ > test_vcpus = 1 > libvirt_rasd_template_changes = 707 > libvirt_rasd_new_changes = 805 > +libvirt_ac_diskpool_changes = 839 > > def setup_env(server, virt="Xen"): > status = PASS > @@ -227,8 +228,11 @@ > if curr_cim_rev >= libvirt_rasd_new_changes: > exp_len = 16 > if virt == 'KVM': > - if curr_cim_rev >= libvirt_rasd_new_changes: > + if curr_cim_rev >= libvirt_rasd_new_changes and \ > + curr_cim_rev < libvirt_ac_diskpool_changes: > exp_len = 8 > + else: > + exp_len = 56 > The number of DiskRASDs returned will depend on the number of images in the storage pool. Instead of hardcoding this number, you'll need to query libvirt to see how many volumes are in the pool.

1 0