This patch migrates the S3A connector to use the V2 AWS SDK.
This is a significant change at the source code level.
Any applications using the internal extension/override points in
the filesystem connector are likely to break.
This includes but is not limited to:
- Code invoking methods on the S3AFileSystem class
which used classes from the V1 SDK.
- The ability to define the factory for the `AmazonS3` client, and
to retrieve it from the S3AFileSystem. There is a new factory
API and a special interface S3AInternals to access a limited
set of internal classes and operations.
- Delegation token and auditing extensions.
- Classes trying to integrate with the AWS SDK.
All standard V1 credential providers listed in the option
fs.s3a.aws.credentials.provider will be automatically remapped to their
V2 equivalent.
Other V1 Credential Providers are supported, but only if the V1 SDK is
added back to the classpath.
The SDK Signing plugin has changed; all v1 signers are incompatible.
There is no support for the S3 "v2" signing algorithm.
Finally, the aws-sdk-bundle JAR has been replaced by the shaded V2
equivalent, "bundle.jar", which is now exported by the hadoop-aws module.
Consult the document aws_sdk_upgrade for the full details.
Contributed by Ahmar Suhail + some bits by Steve Loughran
As well as the POM update, this patch moves to the (renamed) verify methods.
Backporting mockito test changes may now require cherrypicking this patch, otherwise
use the old method names.
Contributed by Anmol Asrani
To avoid the ABFS instance getting closed due to GC while the streams are working, attach the ABFS instance to a backReference opaque object and passing down to the streams so that we have a hard reference while the streams are working.
Contributed by: Mehakmeet Singh
* Add jdiff xml files from 3.3.6 release.
* Declare 3.3.6 as the latest stable release.
* Copy release notes.
(cherry picked from commit 7db9895000)
(cherry picked from commit cc121e2124aa01458dc296a060edc5e21a295268)
This modifies the manifest committer so that the list of files
to rename is passed between stages as a file of
writeable entries on the local filesystem.
The map of directories to create is still passed in memory;
this map is built across all tasks, so even if many tasks
created files, if they all write into the same set of directories
the memory needed is O(directories) with the
task count not a factor.
The _SUCCESS file reports on heap size through gauges.
This should give a warning if there are problems.
Contributed by Steve Loughran
This:
1. Adds optLong, optDouble, mustLong and mustDouble
methods to the FSBuilder interface to let callers explicitly
passin long and double arguments.
2. The opt() and must() builder calls which take float/double values
now only set long values instead, so as to avoid problems
related to overloaded methods resulting in a ".0" being appended
to a long value.
3. All of the relevant opt/must calls in the hadoop codebase move to
the new methods
4. And the s3a code is resilient to parse errors in is numeric options
-it will downgrade to the default.
This is nominally incompatible, but the floating-point builder methods
were never used: nothing currently expects floating point numbers.
For anyone who wants to safely set numeric builder options across all compatible
releases, convert the number to a string and then use the opt(String, String)
and must(String, String) methods.
Contributed by Steve Loughran
The HDFS lease APIs have been replicated as interfaces in hadoop-common so other filesystems can
also implement them. Applications which use the leasing APIs should migrate to the new
interface where possible.
Contributed by Stephen Wu
The log level can only be set on Log4J log implementations;
probes are used to downgrade to a warning when other
logging back ends are used
Contributed by Viraj Jasani
Even though DiskChecker.mkdirsWithExistsCheck() will create the directory tree,
it is only called *after* the enumeration of directories with available
space has completed.
Directories which don't exist are reported as having 0 space, therefore
the mkdirs code is never reached.
Adding a simple mkdirs() -without bothering to check the outcome-
ensures that if a dir has been deleted then it will be reconstructed
if possible. If it can't it will still have 0 bytes of space
reported and so be excluded from the allocation.
Contributed by Steve Loughran
Expands on the comments in cluster config to tell people
they shouldn't be running a cluster without a private VLAN
in cloud, that Knox is good here, and unsecured clusters
without a VLAN are just computation-as-a-service to crypto miners
Contributed by Steve Loughran
Changes method name of RPC.Builder#setnumReaders to setNumReaders()
The original method is still there, just marked deprecated.
It is the one which should be used when working with older branches.
Contributed by Haiyang Hu
When closing we need to wrap the flush() in a try .. finally, otherwise
when flush throws it will stop completion of the remainder of the
close activities and in particular the close of the underlying wrapped
stream object resulting in a resource leak.
Contributed by Colm Dougan
Contributed by Viraj Jasani <vjasani@apache.org>
Signed-off-by: Chris Nauroth <cnauroth@apache.org>
Signed-off-by: Steve Loughran <stevel@apache.org>
Signed-off-by: Mingliang Liu <liuml07@apache.org>
Part of HADOOP-18469 and the hardening of XML/XSL parsers.
Followup to the main HADOOP-18575 patch, to improve performance when
working with xml/xsl engines which don't support the relevant attributes.
Include this change when backporting.
Contributed by PJ Fanning.
The kerberos RPC does not declare any restriction on
characters used in kerberos names, though
implementations MAY be more restrictive.
If the kerberos controller supports use non-conventional
principal names *and the kerberos admin chooses to use them*
this can confuse some of the parsing.
The obvious solution is for the enterprise admins to "not do that"
as a lot of things break, bits of hadoop included.
Harden the hadoop code slightly so at least we fail more gracefully,
so people can then get in touch with their sysadmin and tell them
to stop it.
This has triggered an OOM in a process which was churning through s3a fs
instances; the increased memory footprint of IOStatistics amplified what
must have been a long-standing issue with FS instances being created
and not closed()
* Makes sure instrumentation is closed when the FS is closed.
* Uses a weak reference from metrics to instrumentation, so even
if the FS wasn't closed (see HADOOP-18478), this back reference
would not cause the S3AInstrumentation reference to be retained.
* If S3AFileSystem is configured to log at TRACE it will log the
calling stack of initialize(), so help identify where the
instance is being created. This should help track down
the cause of instance leakage.
Contributed by Steve Loughran.
This addresses HADOOP-18521, "ABFS ReadBufferManager buffer sharing
across concurrent HTTP requests" by not trying to cancel
in progress reads.
It supercedes HADOOP-18528, which disables the prefetching.
If that patch is applied *after* this one, prefetching
will be disabled.
As well as changing the default value in the code,
core-default.xml is updated to set
fs.azure.enable.readahead = true
As a result, if Configuration.get("fs.azure.enable.readahead")
returns a non-null value, then it can be inferred that
it was set in or core-default.xml (the fix is present)
or in core-site.xml (someone asked for it).
Contributed by Pranav Saxena.
* Exactly 1 sending thread per an RPC connection.
* If the calling thread is interrupted before the socket write, it will be skipped instead of sending it anyways.
* If the calling thread is interrupted during the socket write, the write will finish.
* RPC requests will be written to the socket in the order received.
* Sending thread is only started by the receiving thread.
* The sending thread periodically checks the shouldCloseConnection flag.
Disables block prefetching on ABFS InputStreams, by setting
fs.azure.enable.readahead to false in core-default.xml and
the matching java constant.
This prevents
HADOOP-18521. ABFS ReadBufferManager buffer sharing across concurrent HTTP requests.
Once a fix for that is committed, this change can be reverted.
Contributed by Mehakmeet Singh.
Move construction of XML parsers in YARN
modules to using the locked-down parser factory
of HADOOP-18469.
One exception: GpuDeviceInformationParser still supports DTD resolution;
all other features are disabled.
Contributed by P J Fanning
Add to XMLUtils a set of methods to create secure XML Parsers/transformers, locking down DTD, schema, XXE exposure.
Use these wherever XML parsers are created.
Contributed by PJ Fanning
Make S3APrefetchingInputStream.seek() completely lazy. Calls to seek() will not affect the current buffer nor interfere with prefetching, until read() is called.
This change allows various usage patterns to benefit from prefetching, e.g. when calling readFully(position, buffer) in a loop for contiguous positions the intermediate internal calls to seek() will be noops and prefetching will have the same performance as in a sequential read.
Contributed by Alessandro Passaro.
part of HADOOP-18103.
Also introducing a config fs.s3a.vectored.active.ranged.reads
to configure the maximum number of number of range reads a
single input stream can have active (downloading, or queued)
to the central FileSystem instance's pool of queued operations.
This stops a single stream overloading the shared thread pool.
Contributed by: Mukund Thakur
This problem surfaced in impala integration tests
IMPALA-11592. TestLocalCatalogRetries.test_fetch_metadata_retry fails in S3 build
after the change
HADOOP-17461. Add thread-level IOStatistics Context
The actual GC race condition came with
HADOOP-18091. S3A auditing leaks memory through ThreadLocal references
The fix for this is, if our hypothesis is correct, in WeakReferenceMap.create()
where a strong reference to the new value is kept in a local variable
*and referred to later* so that the JVM will not GC it.
Along with the fix, extra assertions ensure that if the problem is not fixed,
applications will fail faster/more meaningfully.
Contributed by Steve Loughran.
part of HADOOP-18103.
While merging the ranges in CheckSumFs, they are rounded up based on the
value of checksum bytes size which leads to some ranges crossing the EOF
thus they need to be fixed else it will cause EOFException during actual reads.
Contributed By: Mukund Thakur
