It seems that one and the same problem presents in both Hortonworks SandPit virtual machine and HDInsight Emulator (most likely because of old versions of Hadoop): jobs are always in pending status and no explicit errors/exceptions appear in jobtracker/tasktracker logs.
The solution for the Emulator is to include the next section into the Hadoop configuration file:
<property> <name>mapreduce.tasktracker.reserved.physicalmemory.mb</name> <value>2048</value> <description> Maximum physical memory tasktracker should reserve for mapreduce tasks. If tasks use more than the limit, task using maximum memory will be killed. Expert only: Set this value if tasktracker should use a certain amount of memory for mapreduce tasks. In MapR Distro warden figures this number based on services configured on a node. Setting mapreduce.tasktracker.reserved.physicalmemory.mb to -1 will disable physical memory accounting and task management. </description> </property>
For emulator-based installation the config is stored here: C:\Hadoop\hadoop-1.1.0-SNAPSHOT\conf\mapred-site.xml.