set 'hive.query.name' ... View more

Check /tmp/hive for hive log and you will see the error stacktrace there. ... View more

There is no way to run multiple DAG concurrently in single Tez AM, and this is Tez limitation. But HiveServer2 support multiple concurrent queries with multiple Tez AM. ... View more

It looks like Tez AM container cannot be scheduled. Typically this is caused by not enough resource. Please check your cluster capacity to ensure you have enough resource for Tez container. ... View more

Increasing timeout cannot guarantee tez dags to be loaded. You may want to find why tez dags won't load also. ... View more

This is not the app log. Please use "yarn log --applicationId <appId>" to retrieve the app log. ... View more

Map vertex get killed because Reduce vertex failed. Any log for reduce vertex? ... View more

Your resource is limited. To solve the problem, either you allocate more resource or just request ask fro resource. ... View more

Never mind about cluster size. I saw you are using sandbox. Probably job cannot be started because there is not enough resource. You can try allocate more memory for VM or decrease tez container size. ... View more

Cluster size: yarn node -list Hive log: /tmp/<username>/hive.log ... View more

Job is not started at all. Please share more info for debugging(cluster size, query, data size, hive log, etc) ... View more

Typically tez.runtime.framework.local.dirs isn't used by user, and it's only used internally by Tez. Its value comes from YARN local dirs configuration through environment variable for containers. There are several ways for LocalDirAllocator to find no valid local dir: 1. disk/dir is full; 2. disk/dir is not writable; 3. local dirs settings is not correct. Since mapreduce based jobs work fine for you, I assume there is not issue about disk being full or unwritable. You should check whether env var LOCAL_DIRS is correct by looking at container launching script. ... View more

You can set hive/tez to use a specified queue. For tez, the configuration is 'tez.queu.name'. Regarding to queue capacity settings, please refer to this link. ... View more

Looks a bug in either Hive or Tez. Would you mind filing a jira in apache and uploading logs there? ... View more

Thanks for the performance evaluation! Any explanation why Databricks CSV library's performance performs well? ... View more

in hive cli ... View more

This log is incomplete and doesn't help. Would you mind sharing the hive log when you trigger the same error? Your can share it via pastebin or collabedit. ... View more