Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

jj #3

Merged
merged 519 commits into from
May 23, 2020
Merged

jj #3

merged 519 commits into from
May 23, 2020

Conversation

tooptoop4
Copy link
Owner

No description provided.

findepi and others added 30 commits April 27, 2020 21:12
The visitor always returned `null` result, and using `Node` as result
type was misleading.
Default implementation provided by `DefaultTraversalVisitor` is suitable
for `Void` result. Moreover, `DefaultTraversalVisitor` is useful base
class for when traversing a tree (a `Void` result), not when
transforming it (non-`Void` result).
The decision of how unnest expressions map to output fields
is now determined by the analyzer. Doing this in the planner
duplicates effort and is brittle.
This simplifies join operator as:
1. Probe is spilled in just one place. Previously probe
   was spilled when page was added or output was requested
2. Join operator state is now split between two internal WorkProcessors
   which a) join probe b) manage final unspilling
3. Operator state is not managed across multiple classic in/out operator methods
It's caller responsibility to choose when we append outer row.
Stress testing showed that Rubix caching is not stable with parallel
warmup enabled. Temporarily diabling by default.
In our testing, a cloud's network proved to be not reliable. We observed
data corruption when transmitting data over TCP between Presto nodes
(internal communication unsecured, no compression).

Verify data integrity to prevent incorrect query results.

Optionally retry when data corruption is detected.
sopel39 and others added 29 commits May 21, 2020 14:32
Internal Rubix configuration is modified by Rubix (BookKeeper
and LocalDataTransferServer). If same cache key is used then
FileSystem for such configuration can be cached by PrestoFileSystemCache
causing cache to be disabled.
Co-authored-by: qqibrow <qqibrow@gmail.com>
Co-authored-by: Zhenxiao Luo <luoz@uber.com>
This allows running queries over the results of a raw Elasticsearch query.
It extends the syntax of the enhanced ES table names with the following:

    SELECT * FROM es.default."<index>$query:<base32-encoded ES query>"

The query is base32-encoded to avoid having to deal with escaping quotes and case
sensitivity issues in table identifiers.

The result of these query tables is a table with a single row and a single column
named "result" of type JSON.
We can directly use the stream method in any collection.
In real deployments Rubix on coordinator is not
part of caching node pool. This pollutes coordinator
log with Rubix error messages whenever coordinator reads
a file (e.g ACID version file). This commit disables
caching on coordinator by default. Caching on coordinator
can still be enabled via feature toggle for development
purpose.
Presto's JSON type imposes additional constraints that are not
desirable for this use case: JSON values must be equatable and
orderable. This requires parsing and re-organizing the document
to canonicalize field ordering.

Using VARCHAR will also play better with the SQL 2016 JSON features,
which operate on binary or string data directly.
Additionally
- remove spfileXE.ora and set paramaters programmatically
- remove redundant "super." in TestingOracleServer
@tooptoop4 tooptoop4 merged commit 9f1f415 into tooptoop4:master May 23, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.