qemu: Recompute downtime and total time when migration completes

Total time of a migration and total downtime transfered from a source to
a destination host do not count with the transfer time to the
destination host and with the time elapsed before guest CPUs are
resumed. Thus, source libvirtd remembers when migration started and when
guest CPUs were paused. Both timestamps are transferred to destination
libvirtd which uses them to compute total migration time and total
downtime. Obviously, this requires the time to be synchronized between
the two hosts. The reported times are useless otherwise but they would
be equally useless if we didn't do this recomputation so don't lose
anything by doing it.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>

diff --git a/src/libvirt.c b/src/libvirt.c
index 00b4d6f14cd33b9bf424a1908c17e9517b0febcd..941c51896b539d350c4979553cd359479ae4d2ff 100644 (file)
--- a/src/libvirt.c
+++ b/src/libvirt.c
@@ -17581,7 +17581,10 @@ virDomainGetJobInfo(virDomainPtr domain, virDomainJobInfoPtr info)
  * return statistics about a recently completed job. Specifically, this
  * flag may be used to query statistics of a completed incoming migration.
  * Statistics of a completed job are automatically destroyed once read or
- * when libvirtd is restarted.
+ * when libvirtd is restarted. Note that time information returned for
+ * completed migrations may be completely irrelevant unless both source and
+ * destination hosts have synchronized time (i.e., NTP daemon is running on
+ * both of them).
  *
  * Returns 0 in case of success and -1 in case of failure.
  */

diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
index 78d6e8c21b51eeb0ab5bf59ea6982abc0723ff89..c0306d7f28a5ca4cdde5d91db0c23795f143f54e 100644 (file)
--- a/src/qemu/qemu_domain.c
+++ b/src/qemu/qemu_domain.c
@@ -222,10 +222,38 @@ qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
     if (virTimeMillisNow(&now) < 0)
         return -1;
 
+    if (now < jobInfo->started) {
+        VIR_WARN("Async job starts in the future");
+        jobInfo->started = 0;
+        return 0;
+    }
+
     jobInfo->timeElapsed = now - jobInfo->started;
     return 0;
 }
 
+int
+qemuDomainJobInfoUpdateDowntime(qemuDomainJobInfoPtr jobInfo)
+{
+    unsigned long long now;
+
+    if (!jobInfo->stopped)
+        return 0;
+
+    if (virTimeMillisNow(&now) < 0)
+        return -1;
+
+    if (now < jobInfo->stopped) {
+        VIR_WARN("Guest's CPUs stopped in the future");
+        jobInfo->stopped = 0;
+        return 0;
+    }
+
+    jobInfo->status.downtime = now - jobInfo->stopped;
+    jobInfo->status.downtime_set = true;
+    return 0;
+}
+
 int
 qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                         virDomainJobInfoPtr info)

diff --git a/src/qemu/qemu_domain.h b/src/qemu/qemu_domain.h
index 2e16cde10e31da43b1294122c141be0b8b2f7bf3..479b51fe8be374b9f65ce02eeb61e57745f81449 100644 (file)
--- a/src/qemu/qemu_domain.h
+++ b/src/qemu/qemu_domain.h
@@ -105,6 +105,7 @@ typedef qemuDomainJobInfo *qemuDomainJobInfoPtr;
 struct _qemuDomainJobInfo {
     virDomainJobType type;
     unsigned long long started; /* When the async job started */
+    unsigned long long stopped; /* When the domain's CPUs were stopped */
     /* Computed values */
     unsigned long long timeElapsed;
     unsigned long long timeRemaining;
@@ -391,6 +392,8 @@ bool qemuDomainAgentAvailable(qemuDomainObjPrivatePtr priv,
 
 int qemuDomainJobInfoUpdateTime(qemuDomainJobInfoPtr jobInfo)
     ATTRIBUTE_NONNULL(1);
+int qemuDomainJobInfoUpdateDowntime(qemuDomainJobInfoPtr jobInfo)
+    ATTRIBUTE_NONNULL(1);
 int qemuDomainJobInfoToInfo(qemuDomainJobInfoPtr jobInfo,
                             virDomainJobInfoPtr info)
     ATTRIBUTE_NONNULL(1) ATTRIBUTE_NONNULL(2);

diff --git a/src/qemu/qemu_migration.c b/src/qemu/qemu_migration.c
index 0a002951a3ed86e05c35684d2abe985126e04d59..801e5450f2c2f350180d8b8393f28b2f704ebe81 100644 (file)
--- a/src/qemu/qemu_migration.c
+++ b/src/qemu/qemu_migration.c
@@ -623,6 +623,9 @@ qemuMigrationCookieStatisticsXMLFormat(virBufferPtr buf,
     virBufferAddLit(buf, "<statistics>\n");
     virBufferAdjustIndent(buf, 2);
 
+    virBufferAsprintf(buf, "<started>%llu</started>\n", jobInfo->started);
+    virBufferAsprintf(buf, "<stopped>%llu</stopped>\n", jobInfo->stopped);
+
     virBufferAsprintf(buf, "<%1$s>%2$llu</%1$s>\n",
                       VIR_DOMAIN_JOB_TIME_ELAPSED,
                       jobInfo->timeElapsed);
@@ -891,6 +894,9 @@ qemuMigrationCookieStatisticsXMLParse(xmlXPathContextPtr ctxt)
     status = &jobInfo->status;
     jobInfo->type = VIR_DOMAIN_JOB_COMPLETED;
 
+    virXPathULongLong("string(./started[1])", ctxt, &jobInfo->started);
+    virXPathULongLong("string(./stopped[1])", ctxt, &jobInfo->stopped);
+
     virXPathULongLong("string(./" VIR_DOMAIN_JOB_TIME_ELAPSED "[1])",
                       ctxt, &jobInfo->timeElapsed);
     virXPathULongLong("string(./" VIR_DOMAIN_JOB_TIME_REMAINING "[1])",
@@ -2015,6 +2021,7 @@ qemuMigrationWaitForCompletion(virQEMUDriverPtr driver,
     }
 
     if (jobInfo->type == VIR_DOMAIN_JOB_COMPLETED) {
+        qemuDomainJobInfoUpdateDowntime(jobInfo);
         VIR_FREE(priv->job.completed);
         if (VIR_ALLOC(priv->job.completed) == 0)
             *priv->job.completed = *jobInfo;
@@ -3597,8 +3604,10 @@ qemuMigrationRun(virQEMUDriverPtr driver,
         VIR_FORCE_CLOSE(fd);
     }
 
-    if (priv->job.completed)
+    if (priv->job.completed) {
         qemuDomainJobInfoUpdateTime(priv->job.completed);
+        qemuDomainJobInfoUpdateDowntime(priv->job.completed);
+    }
 
     cookieFlags |= QEMU_MIGRATION_COOKIE_NETWORK |
                    QEMU_MIGRATION_COOKIE_STATS;
@@ -4811,6 +4820,10 @@ qemuMigrationFinish(virQEMUDriverPtr driver,
                 }
                 goto endjob;
             }
+            if (priv->job.completed) {
+                qemuDomainJobInfoUpdateTime(priv->job.completed);
+                qemuDomainJobInfoUpdateDowntime(priv->job.completed);
+            }
         }
 
         dom = virGetDomain(dconn, vm->def->name, vm->def->uuid);

diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c
index b1d8a324dd3866bfa43eb1932ed3d77124256ead..c70290af31d3bca558e54c7ae70daa8d7d8a8b81 100644 (file)
--- a/src/qemu/qemu_process.c
+++ b/src/qemu/qemu_process.c
@@ -755,6 +755,9 @@ qemuProcessHandleStop(qemuMonitorPtr mon ATTRIBUTE_UNUSED,
         VIR_DEBUG("Transitioned guest %s to paused state",
                   vm->def->name);
 
+        if (priv->job.current)
+            ignore_value(virTimeMillisNow(&priv->job.current->stopped));
+
         virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_UNKNOWN);
         event = virDomainEventLifecycleNewFromObj(vm,
                                          VIR_DOMAIN_EVENT_SUSPENDED,
@@ -2889,7 +2892,8 @@ qemuProcessStartCPUs(virQEMUDriverPtr driver, virDomainObjPtr vm,
 }
 
 
-int qemuProcessStopCPUs(virQEMUDriverPtr driver, virDomainObjPtr vm,
+int qemuProcessStopCPUs(virQEMUDriverPtr driver,
+                        virDomainObjPtr vm,
                         virDomainPausedReason reason,
                         qemuDomainAsyncJob asyncJob)
 {
@@ -2907,6 +2911,9 @@ int qemuProcessStopCPUs(virQEMUDriverPtr driver, virDomainObjPtr vm,
     if (ret < 0)
         goto cleanup;
 
+    if (priv->job.current)
+        ignore_value(virTimeMillisNow(&priv->job.current->stopped));
+
     virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, reason);
     if (virDomainLockProcessPause(driver->lockManager, vm, &priv->lockState) < 0)
         VIR_WARN("Unable to release lease on %s", vm->def->name);

diff --git a/tools/virsh.pod b/tools/virsh.pod
index 44110275cd8baac2b66e2690bc797dd75aadc63f..60ee5152f955b6f9af0c5cf83333584536b554e8 100644 (file)
--- a/tools/virsh.pod
+++ b/tools/virsh.pod
@@ -1136,7 +1136,10 @@ Abort the currently running domain job.
 Returns information about jobs running on a domain. I<--completed> tells
 virsh to return information about a recently finished job. Statistics of
 a completed job are automatically destroyed once read or when libvirtd
-is restarted.
+is restarted. Note that time information returned for completed
+migrations may be completely irrelevant unless both source and
+destination hosts have synchronized time (i.e., NTP daemon is running
+on both of them).
 
 =item B<domname> I<domain-id-or-uuid>