[
https://issues.apache.org/jira/browse/DRILL-5716?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16151221#comment-16151221
]
ASF GitHub Bot commented on DRILL-5716:
---------------------------------------
Github user paul-rogers commented on a diff in the pull request:
https://github.com/apache/drill/pull/928#discussion_r136671513
--- Diff: exec/java-exec/src/main/resources/drill-module.conf ---
@@ -344,6 +357,20 @@ drill.exec: {
# Full workspace name should be indicated (including schema and
workspace separated by dot).
# Workspace MUST be file-based and writable. Workspace name is
case-sensitive.
default_temporary_workspace: "dfs.tmp"
+
+ // Resource management
+ rm : {
+ // Memory per node normally comes from the direct memory alloated on
the JVM
+ // command line. This parameter, if other than 0, further limits the
amount.
+ // Primarily for testing.
+ memory_per_node: 0,
+ // The number of available CPUs normally comes directly from the
system itself.
+ // This parameter, if other than 0, further limits the number of CPUs
will
+ // will consider when planning. Note that, sadly, this parameter does
not
+ // limit *actual* CPU usage; only the amount of CPU assumed to exist
when
+ // planning and managing queries. Primarily for testing.
+ cpus_per_node: 0,
+ }
--- End diff --
Well, it is not for testing only. Suppose that Drill runs under YARN. Then
Drill should not use the full set of CPUs; it should use the number of vcores
granted by YARN. In this case, the YARN launcher can use
`-Ddrill.exec.rm.cpus_per_node=5` to specify the number of vcores.
The memory per node is a safety-valve. At present, several operators (at
least three) decline to participate in the memory assignment scheme. We may
find it necessary to artificially limit memory in order to force certain
queries to work if they end up using far more memory than expected. (True, the
memory ratio runtime option also solves this issue...)
These options do not appear in `drill-override-example.conf`, so users
won't know about them.
> Queue-based memory assignment for buffering operators
> -----------------------------------------------------
>
> Key: DRILL-5716
> URL: https://issues.apache.org/jira/browse/DRILL-5716
> Project: Apache Drill
> Issue Type: Improvement
> Affects Versions: 1.12.0
> Reporter: Paul Rogers
> Assignee: Paul Rogers
>
> Apache Drill already has a queueing feature based on ZK semaphores. We did a
> bit of testing to show that the feature does, in fact work. We propose to
> enhance the feature with some light revisions to make work with the "managed"
> external sort and the newly-added spilling feature for the hash agg operator.
> The key requirement is to build on what we have for now; we may want to
> tackle a larger project to create a more complete solution later.
> Existing functionality:
> * Two ZK-based queues called the “small” and “large” query queues.
> * A threshold, call it T, given as a query cost, to determine the queue into
> which a query will go.
> * Admit levels for the two queues: call them Qs and Ql.
> Basically, when a query comes in:
> * Plan the query as usual.
> * Obtain the final query cost from the planner, call this C.
> * If C<T, the query goes into the small queue, else it goes into the large
> queue.
> * Suppose the small queue. Ask ZK if the query can run.
> * ZK checks if Qs queries are already running. If so, the query waits, else
> the query runs.
> The proposed changes include:
> * Refactor the code to provide a queueing API that supports a variety of
> queuing mechanisms.
> * Provide three: the null queue (default), an in-process queue (for testing)
> and the ZK queues.
> * Modify the query profile web UI to show two new bits of information about
> queues:
> - The queue to which the query was sent.
> - The total planning cost.
> * Modify the query profile web UI to show two memory assignment numbers:
> - Total memory allocated to the query
> - Memory per sort or hash-add operator
> Then, add to the queue mechanism the ability to do memory assignment:
> * Provide a weight, W: every small query gets 1 unit, every large query gets
> W units.
> * Use the queue admit levels to determine total units: U = Qs + W * Ql.
> * Obtain total direct memory from the system. M.
> * Subtract a reserve percent R for overhead.
> * Do the math to get the memory per query for each query:
> * For the small queue: (M - R) / U
> * For the large queue: (M - R) / U * W
> * Use this memory amount as the “memory per query” number in the existing
> sort/hash-agg memory assignment (instead of the fixed 2 GB.)
> The result will be a nice incremental addition to what we already have, and
> should make it a bit easier people to actually use the feature (because they
> can see the planning numbers and see the queues used, allowing them to
> effectively tune the system.)
> The API used for the above features also allow third parties to add on a more
> robust admission control feature as needed, perhaps tying into an existing
> queueing mechanism of their choice.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
