--- /dev/null
+Inside the virt plugin
+======================
+
+Originally written: 20161111
+
+Last updated: 20161124
+
+This document will explain the new domain tag support introduced
+in the virt plugin, and will provide one important use case for this feature.
+In the reminder of this document, we consider
+
+* libvirt <= 2.0.0
+* QEMU <= 2.6.0
+
+
+Domain tags and domains partitioning across virt reader instances
+-----------------------------------------------------------------
+
+The virt plugin gained the `Instances` option. It allows to start
+more than one reader instance, so the the libvirt domains could be queried
+by more than one reader thread.
+The default value for `Instances` is `1`.
+With default settings, the plugin will behave in a fully transparent,
+backward compatible way.
+It is recommended to set this value to one multiple of the
+daemon `ReadThreads` value.
+
+Each reader instance will query only a subset of the libvirt domain.
+The subset is identified as follows:
+
+1. Each virt reader instance is named `virt-$NUM`, where `NUM` is
+ the progressive order of instances. If you configure `Instances 3`
+ you will have `virt-0`, `virt-1`, `virt-2`. Please note: the `virt-0`
+ instance is special, and will always be available.
+2. Each virt reader instance will iterate over all the libvirt active domains,
+ and will look for one `tag` attribute (see below) in the domain metadata section.
+3. Each virt reader instance will take care *only* of the libvirt domains whose
+ tag matches with its own
+4. The special `virt-0` instance will take care of all the libvirt domains with
+ no tags, or with tags which are not in the set \[virt-0 ... virt-$NUM\]
+
+Collectd will just use the domain tags, but never enforces or requires them.
+It is up to an external entity, like a software management system,
+to attach and manage the tags to the domain.
+
+Please note that unless you have such tag-aware management sofware,
+it most likely make no sense to enable more than one reader instance on your
+setup.
+
+
+Libvirt tag metadata format
+----------------------------
+
+This is the snipped to be added to libvirt domains:
+
+ <ovirtmap:tag xmlns:ovirtmap="http://ovirt.org/ovirtmap/tag/1.0">
+ $TAG
+ </ovirtmap:tag>
+
+it must be included in the <metadata> section.
+
+Check the `src/virt_test.c` file for really minimal example of libvirt domains.
+
+
+Examples
+--------
+
+### Example one: 10 libvirt domains named "domain-A" ... "domain-J", virt plugin with instances=5, using 5 different tags
+
+
+ libvirt domain name - tag - read instance - reason
+ domain-A virt-0 0 tag match
+ domain-B virt-1 1 tag match
+ domain-C virt-2 2 tag match
+ domain-D virt-3 3 tag match
+ domain-E virt-4 4 tag match
+ domain-F virt-0 0 tag match
+ domain-G virt-1 1 tag match
+ domain-H virt-2 2 tag match
+ domain-I virt-3 3 tag match
+ domain-J virt-4 4 tag match
+
+
+ Because the domain where properly tagged, all the read instances have even load. Please note that the the virt plugin
+ knows nothing, and should know nothing, about *how* the libvirt domain are tagged. This is entirely up to the
+ management system.
+
+
+Example two: 10 libvirt domains named "domain-A" ... "domain-J", virt plugin with instances=3, using 5 different tags
+
+
+ libvirt domain name - tag - read instance - reason
+ domain-A virt-0 0 tag match
+ domain-B virt-1 1 tag match
+ domain-C virt-2 2 tag match
+ domain-D virt-3 0 adopted by instance #0
+ domain-E virt-4 0 adopted by instance #0
+ domain-F virt-0 0 rag match
+ domain-G virt-1 1 tag match
+ domain-H virt-2 2 tag match
+ domain-I virt-3 0 adopted by instance #0
+ domain-J virt-4 0 adopted by instance #0
+
+
+ In this case we have uneven load, but no domain is ignored.
+
+
+### Example three: 10 libvirt domains named "domain-A" ... "domain-J", virt plugin with instances=5, using 3 different tags
+
+
+ libvirt domain name - tag - read instance - reason
+ domain-A virt-0 0 tag match
+ domain-B virt-1 1 tag match
+ domain-C virt-2 2 tag match
+ domain-D virt-0 0 tag match
+ domain-E virt-1 1 tag match
+ domain-F virt-2 2 tag match
+ domain-G virt-0 0 tag match
+ domain-H virt-1 1 tag match
+ domain-I virt-2 2 tag match
+ domain-J virt-0 0 tag match
+
+
+ Once again we have uneven load and two idle read instances, but besides that no domain is left unmonitored
+
+
+### Example four: 10 libvirt domains named "domain-A" ... "domain-J", virt plugin with instances=5, partial tagging
+
+
+ libvirt domain name - tag - read instance - reason
+ domain-A virt-0 0 tag match
+ domain-B virt-1 1 tag match
+ domain-C virt-2 2 tag match
+ domain-D virt-0 0 tag match
+ domain-E 0 adopted by instance #0
+ domain-F 0 adopted by instance #0
+ domain-G 0 adopted by instance #0
+ domain-H 0 adopted by instance #0
+ domain-I 0 adopted by instance #0
+ domain-J 0 adopted by instance #0
+
+
+The lack of tags causes uneven load, but no domain are unmonitored.
+
+
+Possible extensions - custom tag format
+---------------------------------------
+
+The aformentioned approach relies on fixed tag format, `virt-$N`. The algorithm works fine with any tag, which
+is just one string, compared for equality. However, using custom strings for tags creates the need for a mapping
+between tags and the read instances.
+This mapping needs to be updated as long as domain are created or destroyed, and the virt plugin needs to be
+notified of the changes.
+
+This adds a significant amount of complexity, with little gain with respect to the fixed schema adopted initially.
+For this reason, the introdution of dynamic, custom mapping was not implemented.
+
+
+Dealing with datacenters: libvirt, qemu, shared storage
+-------------------------------------------------------
+
+When used in a datacenter, QEMU is most often configured to use shared storage. This is
+the default configuration of datacenter management solutions like [oVirt](http://www.ovirt.org).
+The actual shared storage could be implemented on top of NFS for small installations, or most likely
+ISCSI or Fiber Channel. The key takeaway is that the storage is accessed over the network,
+not using e.g. the SATA or PCI bus of any given host, so any network issue could cause
+one or more storage operations to delay, or to be lost entirely.
+
+In that case, the userspace process that requested the operation can end up in the D state,
+and become unresponsive, and unkillable.
+
+
+Dealing with unresponsive domains
+---------------------------------
+
+All the above considered, one robust management or monitoring application must deal with the fact that
+the libvirt API can block for a long time, or forever. This is not an issue or a bug of one specific
+API, but it is rather a byproduct of how libvirt and QEMU interact.
+
+Whenever we query more than one VM, we should take care to avoid that one blocked VM prevent other,
+well behaving VMs to be queried. We don't want one rogue VM to disrupt well-behaving VMs.
+Unfortunately, any way we enumerate VMs, either implicitely, using the libvirt bulk stats API,
+or explicitely, listing all libvirt domains and query each one in turn, we may unpredictably encounter
+one unresponsive VM.
+
+There are many possible approaches to deal with this issue. The virt plugin supports
+a simple but effective approach partitioning the domains, as follows.
+
+1. The virt plugin always register one or more `read` callbacks. The `zero` read callback is guaranteed to
+ be always present, so it performs special duties (more details later)
+ Each callback will be named 'virt-$N', where `N` ranges from 0 (zero) to M-1, where M is the number of instances configured.
+ `M` equals to `5` by default, because this is the same default number of threads in the libvirt worker pool.
+2. Each of the read callbacks queries libvirt for the list of all the active domains, and retrieves the libvirt domain metadata.
+ Both of those operations are safe wrt domain blocked in I/O (they affect only the libvirtd daemon).
+3. Each of the read callbacks extracts the `tag` from the domain metadata using a well-known format (see below).
+ Each of the read callbacks discards any domain which has no tag, or whose tag doesn't match with the read callback tag.
+3.a. The read callback tag equals to the read callback name, thus `virt-$N`. Remember that `virt-0` is guaranteed to be
+ always present.
+3.b. Since the `virt-0` reader is always present, it will take care of domains with no tag, or with unrecognized tag.
+ One unrecognized tag is any tag which has not the scheme `virt-$N`.
+4. Each read callback only samples the subset of domains with matching tag. The `virt-0` reader will possibly do more,
+ but worst case the load will be unbalanced, no domain will be left unsampled.
+
+To make this approach work, some entity must attach the tags to the libvirt domains, in such a way that all
+the domains which run on a given host and insist on the same network-based storage share the same tag.
+This minimizes the disruption, because when using the shared storage, if one domain becomes unresponsive because
+of unavailable storage, the most likely thing to happen is that others domain using the same storage will soon become
+unavailable; should the box run other libvirt domains using other network-based storage, they could be monitored
+safely.
+
+In case of [oVirt](http://www.ovirt.org), the aforementioned tagging is performed by the host agent.
+
+Please note that this approach is ineffective if the host completely lose network access to the storage network.
+In this case, however, no recovery is possible and no damage limitation is possible.
+
+Lastly, please note that if the virt plugin is configured with instances=1, it behaves exactly like before.
+
+
+Addendum: high level overview: libvirt client, libvirt daemon, qemu
+--------------------------------------------------------------------
+
+Let's review how the client application (collectd + virt plugin), the libvirtd daemon and the
+QEMU processes interact with each other.
+
+The libvirt daemon talks to QEMU using the JSON QMP protcol over one unix domain socket.
+The details of the protocol are not important now, but the key part is that the protocol
+is a simple request/response, meaning that libvirtd must serialize all the interactions
+with the QEMU monitor, and must protects its endpoint with a lock.
+No out of order request/responses are possible (e.g. no pipelining or async replies).
+This means that if for any reason one QMP request could not be completed, any other caller
+trying to access the QEMU monitor will block until the blocked caller returns.
+
+To retrieve some key informations, most notably about the block device state or the balloon
+device state, the libvirtd daemon *must* use the QMP protocol.
+
+The QEMU core, including the handling of the QMP protocol, is single-threaded.
+All the above combined make it possible for a client to block forever waiting for one QMP
+request, if QEMU itself is blocked. The most likely cause of block is I/O, and this is especially
+true considering how QEMU is used in a datacenter.
+
#include <libxml/parser.h>
#include <libxml/tree.h>
#include <libxml/xpath.h>
+#include <libxml/xpathInternals.h>
/* Plugin name */
#define PLUGIN_NAME "virt"
"PluginInstanceFormat",
+ "Instances",
+
NULL};
#define NR_CONFIG_KEYS ((sizeof config_keys / sizeof config_keys[0]) - 1)
static int ignore_device_match(ignorelist_t *, const char *domname,
const char *devpath);
-/* Actual list of domains found on last refresh. */
-static virDomainPtr *domains = NULL;
-static int nr_domains = 0;
-
-static void free_domains(void);
-static int add_domain(virDomainPtr dom);
-
/* Actual list of block devices found on last refresh. */
struct block_device {
virDomainPtr dom; /* domain */
char *path; /* name of block device */
};
-static struct block_device *block_devices = NULL;
-static int nr_block_devices = 0;
-
-static void free_block_devices(void);
-static int add_block_device(virDomainPtr dom, const char *path);
-
/* Actual list of network interfaces found on last refresh. */
struct interface_device {
virDomainPtr dom; /* domain */
char *number; /* interface device number */
};
-static struct interface_device *interface_devices = NULL;
-static int nr_interface_devices = 0;
+struct lv_read_state {
+ /* Actual list of domains found on last refresh. */
+ virDomainPtr *domains;
+ int nr_domains;
+
+ struct block_device *block_devices;
+ int nr_block_devices;
+
+ struct interface_device *interface_devices;
+ int nr_interface_devices;
+};
+
+static void free_domains(struct lv_read_state *state);
+static int add_domain(struct lv_read_state *state, virDomainPtr dom);
+
+static void free_block_devices(struct lv_read_state *state);
+static int add_block_device(struct lv_read_state *state, virDomainPtr dom,
+ const char *path);
+
+static void free_interface_devices(struct lv_read_state *state);
+static int add_interface_device(struct lv_read_state *state, virDomainPtr dom,
+ const char *path, const char *address,
+ unsigned int number);
-static void free_interface_devices(void);
-static int add_interface_device(virDomainPtr dom, const char *path,
- const char *address, unsigned int number);
+#define METADATA_VM_PARTITION_URI "http://ovirt.org/ovirtmap/tag/1.0"
+#define METADATA_VM_PARTITION_ELEMENT "tag"
+#define METADATA_VM_PARTITION_PREFIX "ovirtmap"
+
+#define BUFFER_MAX_LEN 256
+#define PARTITION_TAG_MAX_LEN 32
+
+struct lv_read_instance {
+ struct lv_read_state read_state;
+ char tag[PARTITION_TAG_MAX_LEN];
+ size_t id;
+};
+
+struct lv_user_data {
+ struct lv_read_instance inst;
+ user_data_t ud;
+};
+
+#define NR_INSTANCES_DEFAULT 1
+#define NR_INSTANCES_MAX 128
+static int nr_instances = NR_INSTANCES_DEFAULT;
+static struct lv_user_data lv_read_user_data[NR_INSTANCES_MAX];
/* HostnameFormat. */
#define HF_MAX_FIELDS 3
/* Time that we last refreshed. */
static time_t last_refresh = (time_t)0;
-static int refresh_lists(void);
+static int refresh_lists(struct lv_read_instance *inst);
/* ERROR(...) macro for virterrors. */
#define VIRT_ERROR(conn, s) \
submit(dom, type, devname, values, STATIC_ARRAY_SIZE(values));
} /* void submit_derive2 */
-static int lv_init(void) {
- if (virInitialize() != 0)
- return -1;
- else
- return 0;
-}
-
static int lv_config(const char *key, const char *value) {
if (virInitialize() != 0)
return 1;
return 0;
}
+ if (strcasecmp(key, "Instances") == 0) {
+ char *eptr = NULL;
+ double val = strtod(value, &eptr);
+
+ if (*eptr != '\0') {
+ ERROR(PLUGIN_NAME " plugin: Invalid value for Instances = '%s'", value);
+ return 1;
+ }
+ if (val <= 0) {
+ ERROR(PLUGIN_NAME " plugin: Instances <= 0 makes no sense.");
+ return 1;
+ }
+ if (val > NR_INSTANCES_MAX) {
+ ERROR(PLUGIN_NAME " plugin: Instances=%f > NR_INSTANCES_MAX=%i"
+ " use a lower setting or recompile the plugin.",
+ val, NR_INSTANCES_MAX);
+ return 1;
+ }
+
+ nr_instances = (int)val;
+ DEBUG(PLUGIN_NAME " plugin: configured %i instances", nr_instances);
+ return 0;
+ }
+
/* Unrecognised option. */
return -1;
}
-static int lv_read(void) {
+static int lv_read(user_data_t *ud) {
time_t t;
+ struct lv_read_instance *inst = NULL;
+ struct lv_read_state *state = NULL;
+
+ if (ud->data == NULL) {
+ ERROR(PLUGIN_NAME " plugin: NULL userdata");
+ return -1;
+ }
+
+ inst = ud->data;
+ state = &inst->read_state;
- if (conn == NULL) {
+ if (inst->id == 0 && conn == NULL) {
/* `conn_string == NULL' is acceptable. */
conn = virConnectOpenReadOnly(conn_string);
if (conn == NULL) {
/* Need to refresh domain or device lists? */
if ((last_refresh == (time_t)0) ||
((interval > 0) && ((last_refresh + interval) <= t))) {
- if (refresh_lists() != 0) {
+ if (refresh_lists(inst) != 0) {
if (conn != NULL)
virConnectClose(conn);
conn = NULL;
#endif
/* Get CPU usage, memory, VCPU usage for each domain. */
- for (int i = 0; i < nr_domains; ++i) {
+ for (int i = 0; i < state->nr_domains; ++i) {
virDomainInfo info;
virVcpuInfoPtr vinfo = NULL;
virDomainMemoryStatPtr minfo = NULL;
int status;
- status = virDomainGetInfo(domains[i], &info);
+ status = virDomainGetInfo(state->domains[i], &info);
if (status != 0) {
ERROR(PLUGIN_NAME " plugin: virDomainGetInfo failed with status %i.",
status);
continue;
}
- cpu_submit(info.cpuTime, domains[i], "virt_cpu_total");
- memory_submit((gauge_t)info.memory * 1024, domains[i]);
+ cpu_submit(info.cpuTime, state->domains[i], "virt_cpu_total");
+ memory_submit((gauge_t)info.memory * 1024, state->domains[i]);
vinfo = malloc(info.nrVirtCpu * sizeof(vinfo[0]));
if (vinfo == NULL) {
continue;
}
- status = virDomainGetVcpus(domains[i], vinfo, info.nrVirtCpu,
+ status = virDomainGetVcpus(state->domains[i], vinfo, info.nrVirtCpu,
/* cpu map = */ NULL, /* cpu map length = */ 0);
if (status < 0) {
ERROR(PLUGIN_NAME " plugin: virDomainGetVcpus failed with status %i.",
}
for (int j = 0; j < info.nrVirtCpu; ++j)
- vcpu_submit(vinfo[j].cpuTime, domains[i], vinfo[j].number, "virt_vcpu");
+ vcpu_submit(vinfo[j].cpuTime, state->domains[i], vinfo[j].number,
+ "virt_vcpu");
sfree(vinfo);
continue;
}
- status =
- virDomainMemoryStats(domains[i], minfo, VIR_DOMAIN_MEMORY_STAT_NR, 0);
+ status = virDomainMemoryStats(state->domains[i], minfo,
+ VIR_DOMAIN_MEMORY_STAT_NR, 0);
if (status < 0) {
ERROR("virt plugin: virDomainMemoryStats failed with status %i.", status);
}
for (int j = 0; j < status; j++) {
- memory_stats_submit((gauge_t)minfo[j].val * 1024, domains[i],
+ memory_stats_submit((gauge_t)minfo[j].val * 1024, state->domains[i],
minfo[j].tag);
}
}
/* Get block device stats for each domain. */
- for (int i = 0; i < nr_block_devices; ++i) {
+ for (int i = 0; i < state->nr_block_devices; ++i) {
struct _virDomainBlockStats stats;
- if (virDomainBlockStats(block_devices[i].dom, block_devices[i].path, &stats,
+ if (virDomainBlockStats(state->block_devices[i].dom,
+ state->block_devices[i].path, &stats,
sizeof stats) != 0)
continue;
char *type_instance = NULL;
if (blockdevice_format_basename && blockdevice_format == source)
- type_instance = strdup(basename(block_devices[i].path));
+ type_instance = strdup(basename(state->block_devices[i].path));
else
- type_instance = strdup(block_devices[i].path);
+ type_instance = strdup(state->block_devices[i].path);
if ((stats.rd_req != -1) && (stats.wr_req != -1))
submit_derive2("disk_ops", (derive_t)stats.rd_req, (derive_t)stats.wr_req,
- block_devices[i].dom, type_instance);
+ state->block_devices[i].dom, type_instance);
if ((stats.rd_bytes != -1) && (stats.wr_bytes != -1))
submit_derive2("disk_octets", (derive_t)stats.rd_bytes,
- (derive_t)stats.wr_bytes, block_devices[i].dom,
+ (derive_t)stats.wr_bytes, state->block_devices[i].dom,
type_instance);
sfree(type_instance);
} /* for (nr_block_devices) */
/* Get interface stats for each domain. */
- for (int i = 0; i < nr_interface_devices; ++i) {
+ for (int i = 0; i < state->nr_interface_devices; ++i) {
struct _virDomainInterfaceStats stats;
char *display_name = NULL;
switch (interface_format) {
case if_address:
- display_name = interface_devices[i].address;
+ display_name = state->interface_devices[i].address;
break;
case if_number:
- display_name = interface_devices[i].number;
+ display_name = state->interface_devices[i].number;
break;
case if_name:
default:
- display_name = interface_devices[i].path;
+ display_name = state->interface_devices[i].path;
}
- if (virDomainInterfaceStats(interface_devices[i].dom,
- interface_devices[i].path, &stats,
+ if (virDomainInterfaceStats(state->interface_devices[i].dom,
+ state->interface_devices[i].path, &stats,
sizeof stats) != 0)
continue;
if ((stats.rx_bytes != -1) && (stats.tx_bytes != -1))
submit_derive2("if_octets", (derive_t)stats.rx_bytes,
- (derive_t)stats.tx_bytes, interface_devices[i].dom,
+ (derive_t)stats.tx_bytes, state->interface_devices[i].dom,
display_name);
if ((stats.rx_packets != -1) && (stats.tx_packets != -1))
submit_derive2("if_packets", (derive_t)stats.rx_packets,
- (derive_t)stats.tx_packets, interface_devices[i].dom,
- display_name);
+ (derive_t)stats.tx_packets,
+ state->interface_devices[i].dom, display_name);
if ((stats.rx_errs != -1) && (stats.tx_errs != -1))
submit_derive2("if_errors", (derive_t)stats.rx_errs,
- (derive_t)stats.tx_errs, interface_devices[i].dom,
+ (derive_t)stats.tx_errs, state->interface_devices[i].dom,
display_name);
if ((stats.rx_drop != -1) && (stats.tx_drop != -1))
submit_derive2("if_dropped", (derive_t)stats.rx_drop,
- (derive_t)stats.tx_drop, interface_devices[i].dom,
+ (derive_t)stats.tx_drop, state->interface_devices[i].dom,
display_name);
} /* for (nr_interface_devices) */
return 0;
}
-static int refresh_lists(void) {
+static int lv_init_instance(size_t i, plugin_read_cb callback) {
+ struct lv_user_data *lv_ud = &(lv_read_user_data[i]);
+ struct lv_read_instance *inst = &(lv_ud->inst);
+
+ memset(lv_ud, 0, sizeof(*lv_ud));
+
+ ssnprintf(inst->tag, sizeof(inst->tag), "%s-%zu", PLUGIN_NAME, i);
+ inst->id = i;
+
+ user_data_t *ud = &(lv_ud->ud);
+ ud->data = inst;
+ ud->free_func = NULL;
+
+ INFO(PLUGIN_NAME " plugin: reader %s initialized", inst->tag);
+ return plugin_register_complex_read(NULL, inst->tag, callback, 0, ud);
+}
+
+static void lv_clean_read_state(struct lv_read_state *state) {
+ free_block_devices(state);
+ free_interface_devices(state);
+ free_domains(state);
+}
+
+static void lv_fini_instance(size_t i) {
+ struct lv_read_instance *inst = &(lv_read_user_data[i].inst);
+ struct lv_read_state *state = &(inst->read_state);
+
+ lv_clean_read_state(state);
+ INFO(PLUGIN_NAME " plugin: reader %s finalized", inst->tag);
+}
+
+static int lv_init(void) {
+ if (virInitialize() != 0)
+ return -1;
+
+ DEBUG(PLUGIN_NAME " plugin: starting %i instances", nr_instances);
+
+ for (int i = 0; i < nr_instances; ++i)
+ lv_init_instance(i, lv_read);
+
+ return 0;
+}
+
+static int lv_domain_get_tag(xmlXPathContextPtr xpath_ctx, const char *dom_name,
+ char *dom_tag) {
+ char xpath_str[BUFFER_MAX_LEN] = {'\0'};
+ xmlXPathObjectPtr xpath_obj = NULL;
+ xmlNodePtr xml_node = NULL;
+ int err = -1;
+
+ err = xmlXPathRegisterNs(xpath_ctx,
+ (const xmlChar *)METADATA_VM_PARTITION_PREFIX,
+ (const xmlChar *)METADATA_VM_PARTITION_URI);
+ if (err) {
+ ERROR(PLUGIN_NAME " plugin: xmlXpathRegisterNs(%s, %s) failed on domain %s",
+ METADATA_VM_PARTITION_PREFIX, METADATA_VM_PARTITION_URI, dom_name);
+ goto done;
+ }
+
+ ssnprintf(xpath_str, sizeof(xpath_str), "/domain/metadata/%s:%s/text()",
+ METADATA_VM_PARTITION_PREFIX, METADATA_VM_PARTITION_ELEMENT);
+ xpath_obj = xmlXPathEvalExpression((xmlChar *)xpath_str, xpath_ctx);
+ if (xpath_obj == NULL) {
+ ERROR(PLUGIN_NAME " plugin: xmlXPathEval(%s) failed on domain %s",
+ xpath_str, dom_name);
+ goto done;
+ }
+
+ if (xpath_obj->type != XPATH_NODESET) {
+ ERROR(PLUGIN_NAME " plugin: xmlXPathEval(%s) unexpected return type %d "
+ "(wanted %d) on domain %s",
+ xpath_str, xpath_obj->type, XPATH_NODESET, dom_name);
+ goto done;
+ }
+
+ /*
+ * from now on there is no real error, it's ok if a domain
+ * doesn't have the metadata partition tag.
+ */
+ err = 0;
+
+ if (xpath_obj->nodesetval == NULL || xpath_obj->nodesetval->nodeNr != 1) {
+ DEBUG(PLUGIN_NAME " plugin: xmlXPathEval(%s) return nodeset size=%i "
+ "expected=1 on domain %s",
+ xpath_str,
+ (xpath_obj->nodesetval == NULL) ? 0 : xpath_obj->nodesetval->nodeNr,
+ dom_name);
+ } else {
+ xml_node = xpath_obj->nodesetval->nodeTab[0];
+ sstrncpy(dom_tag, (const char *)xml_node->content, PARTITION_TAG_MAX_LEN);
+ }
+
+done:
+ /* deregister to clean up */
+ err = xmlXPathRegisterNs(xpath_ctx,
+ (const xmlChar *)METADATA_VM_PARTITION_PREFIX, NULL);
+
+ if (xpath_obj)
+ xmlXPathFreeObject(xpath_obj);
+
+ return err;
+}
+
+static int is_known_tag(const char *dom_tag) {
+ for (int i = 0; i < nr_instances; ++i)
+ if (!strcmp(dom_tag, lv_read_user_data[i].inst.tag))
+ return 1;
+ return 0;
+}
+
+static int lv_instance_include_domain(struct lv_read_instance *inst,
+ const char *dom_name,
+ const char *dom_tag) {
+ if ((dom_tag[0] != '\0') && (strcmp(dom_tag, inst->tag) == 0))
+ return 1;
+
+ /* instance#0 will always be there, so it is in charge of extra duties */
+ if (inst->id == 0) {
+ if (dom_tag[0] == '\0' || !is_known_tag(dom_tag)) {
+ DEBUG(PLUGIN_NAME " plugin#%s: refreshing domain %s "
+ "with unknown tag '%s'",
+ inst->tag, dom_name, dom_tag);
+ return 1;
+ }
+ }
+
+ return 0;
+}
+
+static int refresh_lists(struct lv_read_instance *inst) {
+ struct lv_read_state *state = &inst->read_state;
int n;
n = virConnectNumOfDomains(conn);
return -1;
}
+ lv_clean_read_state(state);
+
if (n > 0) {
int *domids;
return -1;
}
- free_block_devices();
- free_interface_devices();
- free_domains();
-
/* Fetch each domain and add it to the list, unless ignore. */
for (int i = 0; i < n; ++i) {
virDomainPtr dom = NULL;
xmlDocPtr xml_doc = NULL;
xmlXPathContextPtr xpath_ctx = NULL;
xmlXPathObjectPtr xpath_obj = NULL;
+ char tag[PARTITION_TAG_MAX_LEN] = {'\0'};
dom = virDomainLookupByID(conn, domids[i]);
if (dom == NULL) {
if (il_domains && ignorelist_match(il_domains, name) != 0)
goto cont;
- if (add_domain(dom) < 0) {
- ERROR(PLUGIN_NAME " plugin: malloc failed.");
- goto cont;
- }
-
/* Get a list of devices for this domain. */
xml = virDomainGetXMLDesc(dom, 0);
if (!xml) {
xpath_ctx = xmlXPathNewContext(xml_doc);
+ if (lv_domain_get_tag(xpath_ctx, name, tag) < 0) {
+ ERROR(PLUGIN_NAME " plugin: lv_domain_get_tag failed.");
+ goto cont;
+ }
+
+ if (!lv_instance_include_domain(inst, name, tag))
+ goto cont;
+
+ if (add_domain(state, dom) < 0) {
+ ERROR(PLUGIN_NAME " plugin: malloc failed.");
+ goto cont;
+ }
+
/* Block devices. */
char *bd_xmlpath = "/domain/devices/disk/target[@dev]";
if (blockdevice_format == source)
ignore_device_match(il_block_devices, name, path) != 0)
goto cont2;
- add_block_device(dom, path);
+ add_block_device(state, dom, path);
cont2:
if (path)
xmlFree(path);
ignore_device_match(il_interface_devices, name, address) != 0))
goto cont3;
- add_interface_device(dom, path, address, j + 1);
+ add_interface_device(state, dom, path, address, j + 1);
cont3:
if (path)
xmlFree(path);
sfree(domids);
}
+ DEBUG(PLUGIN_NAME " plugin#%s: refreshing"
+ " domains=%i block_devices=%i iface_devices=%i",
+ inst->tag, state->nr_domains, state->nr_block_devices,
+ state->nr_interface_devices);
+
return 0;
}
-static void free_domains(void) {
- if (domains) {
- for (int i = 0; i < nr_domains; ++i)
- virDomainFree(domains[i]);
- sfree(domains);
+static void free_domains(struct lv_read_state *state) {
+ if (state->domains) {
+ for (int i = 0; i < state->nr_domains; ++i)
+ virDomainFree(state->domains[i]);
+ sfree(state->domains);
}
- domains = NULL;
- nr_domains = 0;
+ state->domains = NULL;
+ state->nr_domains = 0;
}
-static int add_domain(virDomainPtr dom) {
+static int add_domain(struct lv_read_state *state, virDomainPtr dom) {
virDomainPtr *new_ptr;
- int new_size = sizeof(domains[0]) * (nr_domains + 1);
+ int new_size = sizeof(state->domains[0]) * (state->nr_domains + 1);
- if (domains)
- new_ptr = realloc(domains, new_size);
+ if (state->domains)
+ new_ptr = realloc(state->domains, new_size);
else
new_ptr = malloc(new_size);
if (new_ptr == NULL)
return -1;
- domains = new_ptr;
- domains[nr_domains] = dom;
- return nr_domains++;
+ state->domains = new_ptr;
+ state->domains[state->nr_domains] = dom;
+ return state->nr_domains++;
}
-static void free_block_devices(void) {
- if (block_devices) {
- for (int i = 0; i < nr_block_devices; ++i)
- sfree(block_devices[i].path);
- sfree(block_devices);
+static void free_block_devices(struct lv_read_state *state) {
+ if (state->block_devices) {
+ for (int i = 0; i < state->nr_block_devices; ++i)
+ sfree(state->block_devices[i].path);
+ sfree(state->block_devices);
}
- block_devices = NULL;
- nr_block_devices = 0;
+ state->block_devices = NULL;
+ state->nr_block_devices = 0;
}
-static int add_block_device(virDomainPtr dom, const char *path) {
+static int add_block_device(struct lv_read_state *state, virDomainPtr dom,
+ const char *path) {
struct block_device *new_ptr;
- int new_size = sizeof(block_devices[0]) * (nr_block_devices + 1);
+ int new_size =
+ sizeof(state->block_devices[0]) * (state->nr_block_devices + 1);
char *path_copy;
path_copy = strdup(path);
if (!path_copy)
return -1;
- if (block_devices)
- new_ptr = realloc(block_devices, new_size);
+ if (state->block_devices)
+ new_ptr = realloc(state->block_devices, new_size);
else
new_ptr = malloc(new_size);
sfree(path_copy);
return -1;
}
- block_devices = new_ptr;
- block_devices[nr_block_devices].dom = dom;
- block_devices[nr_block_devices].path = path_copy;
- return nr_block_devices++;
+ state->block_devices = new_ptr;
+ state->block_devices[state->nr_block_devices].dom = dom;
+ state->block_devices[state->nr_block_devices].path = path_copy;
+ return state->nr_block_devices++;
}
-static void free_interface_devices(void) {
- if (interface_devices) {
- for (int i = 0; i < nr_interface_devices; ++i) {
- sfree(interface_devices[i].path);
- sfree(interface_devices[i].address);
- sfree(interface_devices[i].number);
+static void free_interface_devices(struct lv_read_state *state) {
+ if (state->interface_devices) {
+ for (int i = 0; i < state->nr_interface_devices; ++i) {
+ sfree(state->interface_devices[i].path);
+ sfree(state->interface_devices[i].address);
+ sfree(state->interface_devices[i].number);
}
- sfree(interface_devices);
+ sfree(state->interface_devices);
}
- interface_devices = NULL;
- nr_interface_devices = 0;
+ state->interface_devices = NULL;
+ state->nr_interface_devices = 0;
}
-static int add_interface_device(virDomainPtr dom, const char *path,
- const char *address, unsigned int number) {
+static int add_interface_device(struct lv_read_state *state, virDomainPtr dom,
+ const char *path, const char *address,
+ unsigned int number) {
struct interface_device *new_ptr;
- int new_size = sizeof(interface_devices[0]) * (nr_interface_devices + 1);
+ int new_size =
+ sizeof(state->interface_devices[0]) * (state->nr_interface_devices + 1);
char *path_copy, *address_copy, number_string[15];
if ((path == NULL) || (address == NULL))
snprintf(number_string, sizeof(number_string), "interface-%u", number);
- if (interface_devices)
- new_ptr = realloc(interface_devices, new_size);
+ if (state->interface_devices)
+ new_ptr = realloc(state->interface_devices, new_size);
else
new_ptr = malloc(new_size);
sfree(address_copy);
return -1;
}
- interface_devices = new_ptr;
- interface_devices[nr_interface_devices].dom = dom;
- interface_devices[nr_interface_devices].path = path_copy;
- interface_devices[nr_interface_devices].address = address_copy;
- interface_devices[nr_interface_devices].number = strdup(number_string);
- return nr_interface_devices++;
+ state->interface_devices = new_ptr;
+ state->interface_devices[state->nr_interface_devices].dom = dom;
+ state->interface_devices[state->nr_interface_devices].path = path_copy;
+ state->interface_devices[state->nr_interface_devices].address = address_copy;
+ state->interface_devices[state->nr_interface_devices].number =
+ strdup(number_string);
+ return state->nr_interface_devices++;
}
static int ignore_device_match(ignorelist_t *il, const char *domname,
}
static int lv_shutdown(void) {
- free_block_devices();
- free_interface_devices();
- free_domains();
+ for (int i = 0; i < nr_instances; ++i) {
+ lv_fini_instance(i);
+ }
if (conn != NULL)
virConnectClose(conn);
void module_register(void) {
plugin_register_config(PLUGIN_NAME, lv_config, config_keys, NR_CONFIG_KEYS);
plugin_register_init(PLUGIN_NAME, lv_init);
- plugin_register_read(PLUGIN_NAME, lv_read);
plugin_register_shutdown(PLUGIN_NAME, lv_shutdown);
}
--- /dev/null
+/**
+ * collectd - src/virt_test.c
+ * Copyright (C) 2016 Francesco Romani <fromani at redhat.com>
+ * Based on
+ * collectd - src/ceph_test.c
+ * Copyright (C) 2015 Florian octo Forster
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; only version 2 of the License is applicable.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+ *
+ * Authors:
+ * Florian octo Forster <octo at collectd.org>
+ **/
+
+#include "virt.c" /* sic */
+#include "testing.h"
+
+#include <unistd.h>
+
+static const char minimal_xml[] =
+ ""
+ "<?xml version=\"1.0\" encoding=\"utf-8\"?>"
+ "<domain type=\"kvm\" xmlns:ovirt=\"http://ovirt.org/vm/tune/1.0\">"
+ " <metadata/>"
+ "</domain>";
+
+static const char minimal_metadata_xml[] =
+ ""
+ "<?xml version=\"1.0\" encoding=\"utf-8\"?>"
+ "<domain type=\"kvm\" xmlns:ovirt=\"http://ovirt.org/vm/tune/1.0\">"
+ " <metadata>"
+ " <ovirtmap:tag "
+ "xmlns:ovirtmap=\"http://ovirt.org/ovirtmap/tag/1.0\">virt-0</ovirtmap:tag>"
+ " </metadata>"
+ "</domain>";
+
+struct xml_state {
+ xmlDocPtr xml_doc;
+ xmlXPathContextPtr xpath_ctx;
+ xmlXPathObjectPtr xpath_obj;
+ char tag[PARTITION_TAG_MAX_LEN];
+};
+
+static int init_state(struct xml_state *st, const char *xml) {
+ memset(st, 0, sizeof(*st));
+
+ st->xml_doc = xmlReadDoc((const xmlChar *)xml, NULL, NULL, XML_PARSE_NONET);
+ if (st->xml_doc == NULL) {
+ return -1;
+ }
+ st->xpath_ctx = xmlXPathNewContext(st->xml_doc);
+ if (st->xpath_ctx == NULL) {
+ return -1;
+ }
+ return 0;
+}
+
+static void fini_state(struct xml_state *st) {
+ if (st->xpath_ctx) {
+ xmlXPathFreeContext(st->xpath_ctx);
+ st->xpath_ctx = NULL;
+ }
+ if (st->xml_doc) {
+ xmlFreeDoc(st->xml_doc);
+ st->xml_doc = NULL;
+ }
+}
+
+#define TAG "virt-0"
+
+DEF_TEST(lv_domain_get_tag_no_metadata_xml) {
+ int err;
+ struct xml_state st;
+ err = init_state(&st, minimal_xml);
+ EXPECT_EQ_INT(0, err);
+
+ err = lv_domain_get_tag(st.xpath_ctx, "test", st.tag);
+
+ EXPECT_EQ_INT(0, err);
+ EXPECT_EQ_STR("", st.tag);
+
+ fini_state(&st);
+ return 0;
+}
+
+DEF_TEST(lv_domain_get_tag_valid_xml) {
+ int err;
+ struct xml_state st;
+ err = init_state(&st, minimal_metadata_xml);
+ EXPECT_EQ_INT(0, err);
+
+ err = lv_domain_get_tag(st.xpath_ctx, "test", st.tag);
+
+ EXPECT_EQ_INT(0, err);
+ EXPECT_EQ_STR(TAG, st.tag);
+
+ return 0;
+}
+
+DEF_TEST(lv_default_instance_include_domain_without_tag) {
+ struct lv_read_instance *inst = NULL;
+ int ret;
+
+ ret = lv_init_instance(0, lv_read);
+ inst = &(lv_read_user_data[0].inst);
+ EXPECT_EQ_STR("virt-0", inst->tag);
+
+ ret = lv_instance_include_domain(inst, "testing", "");
+ EXPECT_EQ_INT(1, ret);
+
+ lv_fini_instance(0);
+ return 0;
+}
+
+DEF_TEST(lv_regular_instance_skip_domain_without_tag) {
+ struct lv_read_instance *inst = NULL;
+ int ret;
+
+ ret = lv_init_instance(1, lv_read);
+ inst = &(lv_read_user_data[1].inst);
+ EXPECT_EQ_STR("virt-1", inst->tag);
+
+ ret = lv_instance_include_domain(inst, "testing", "");
+ EXPECT_EQ_INT(0, ret);
+
+ lv_fini_instance(0);
+ return 0;
+}
+
+DEF_TEST(lv_include_domain_matching_tags) {
+ struct lv_read_instance *inst = NULL;
+ int ret;
+
+ ret = lv_init_instance(0, lv_read);
+ inst = &(lv_read_user_data[0].inst);
+ EXPECT_EQ_STR("virt-0", inst->tag);
+
+ ret = lv_instance_include_domain(inst, "testing", "virt-0");
+ EXPECT_EQ_INT(1, ret);
+
+ ret = lv_init_instance(1, lv_read);
+ inst = &(lv_read_user_data[1].inst);
+ EXPECT_EQ_STR("virt-1", inst->tag);
+
+ ret = lv_instance_include_domain(inst, "testing", "virt-1");
+ EXPECT_EQ_INT(1, ret);
+
+ lv_fini_instance(0);
+ lv_fini_instance(1);
+ return 0;
+}
+
+DEF_TEST(lv_default_instance_include_domain_with_unknown_tag) {
+ struct lv_read_instance *inst = NULL;
+ int ret;
+
+ ret = lv_init_instance(0, lv_read);
+ inst = &(lv_read_user_data[0].inst);
+ EXPECT_EQ_STR("virt-0", inst->tag);
+
+ ret = lv_instance_include_domain(inst, "testing", "unknownFormat-tag");
+ EXPECT_EQ_INT(1, ret);
+
+ lv_fini_instance(0);
+ return 0;
+}
+
+DEF_TEST(lv_regular_instance_skip_domain_with_unknown_tag) {
+ struct lv_read_instance *inst = NULL;
+ int ret;
+
+ ret = lv_init_instance(1, lv_read);
+ inst = &(lv_read_user_data[1].inst);
+ EXPECT_EQ_STR("virt-1", inst->tag);
+
+ ret = lv_instance_include_domain(inst, "testing", "unknownFormat-tag");
+ EXPECT_EQ_INT(0, ret);
+
+ lv_fini_instance(0);
+ return 0;
+}
+#undef TAG
+
+int main(void) {
+ RUN_TEST(lv_domain_get_tag_no_metadata_xml);
+ RUN_TEST(lv_domain_get_tag_valid_xml);
+
+ RUN_TEST(lv_default_instance_include_domain_without_tag);
+ RUN_TEST(lv_regular_instance_skip_domain_without_tag);
+ RUN_TEST(lv_include_domain_matching_tags);
+ RUN_TEST(lv_default_instance_include_domain_with_unknown_tag);
+ RUN_TEST(lv_regular_instance_skip_domain_with_unknown_tag);
+
+ END_TEST;
+}
+
+/* vim: set sw=2 sts=2 et : */