Recent testing showed that the pytests are not
respecting the log level and format set in
conftest.py's configure_logging(). It is using
the default log level of WARNING and the
default formatter.
The issue is that logging.basicConfig() is only
effective the first time it is called. The code
in lib/python/impala_py_lib/helpers.py does a
call to logging.basicConfig() at the global
level, and conftest.py imports that file. This
renders the call in configure_logging()
ineffective.
To avoid this type of confusion, logging.basicConfig()
should only be called from the main() functions for
libraries. This removes the call in lib/python/impala_py_lib
(as it is not needed for a library without a main function).
It also fixes up various other locations to move the
logging.basicConfig() call to the main() function.
Testing:
- Ran the end to end tests and custom cluster tests
- Confirmed the logging format
- Added an assert in configure_logging() to test that
the INFO log level is applied to the root logger.
Change-Id: I5d91b7f910b3606c50bcba4579179a0bc8c20588
Reviewed-on: http://gerrit.cloudera.org:8080/16679
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Currently, the impala-shell 'profile' command only returns the profile
for the most recent profile attempt. There is no way to get the original
query profile (the profile of the first query attempt that failed) from
the impala-shell.
This patch modifies TGetRuntimeProfileReq and TGetRuntimeProfileResp to
add support for returning both the original and retried profiles for a
retried query. When a query is retried, TGetRuntimeProfileResp currently
contains the profile for the most recent query attempt.
TGetRuntimeProfileReq has a new field called 'include_query_attempts'
and when it is set to true, the TGetRuntimeProfileResp will include all
failed profiles in a new field called failed_profiles /
failed_thrift_profiles.
impala-shell has been modified so the 'profile' command has a new set of
options. The syntax is now:
PROFILE [ALL | LATEST | ORIGINAL]
If 'ALL' is specified, both the latest and original profiles are
printed. If 'LATEST' is specified, only the latest profile is printed.
If 'ORIGINAL' is printed, only the original profile is printed. The
default behavior is equivalent to specifying 'LATEST' (which is the
current behavior before this patch as well).
Support for this has only been added to HS2 given that Beeswax is being
deprecated soon. The new 'profile' options have no affect when the
Beeswax protocol is used.
Most of the code change is in impala-hs2-server and impala-server; a lot
of the GetRuntimeProfile code has been re-factored.
Testing:
* Added new impala-shell tests
* Ran core tests
Change-Id: I89cee02947b311e7bf9c7274f47dfc7214c1bb65
Reviewed-on: http://gerrit.cloudera.org:8080/16406
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Many http servers will not accept an http request that has multiple
copies of the "Host" header. A recent toolchain change patches
Thrift so that will not send the extraneous header (in THttpClient).
This change tests that the duplicate headers are not sent,
TESTING:
Ran all end-to-end tests.
Rewrote an existing Shell test to check that only one "Host" header
is sent.
Change-Id: I82996015d0205923e854dac8bb88604778684c46
Reviewed-on: http://gerrit.cloudera.org:8080/15752
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
This removes LD_LIBRARY_PATH and LD_PRELOAD from the
developer's shell and cleans it up. With the preceding
change, toolchain utilities like clang can be run without
a special LD_LIBRARY_PATH.
This fixes a bug where libjvm.so was registered as a
static instead of a shared library, which adds it to the
RUNPATH variable in the binary, which provides a default
search location that can be overriden by LD_LIBRARY_PATH.
Impala binaries don't have the rpath baked in for some
libraries, including Impala-lzo, libgcc and libstdc++.
, so we still need to set LD_LIBRARY_PATH when running
those. That is solved with wrapper scripts that sets
the environment variables only when invoking those
binaries, e.g. starting a daemon or running a backend
test. I added three scripts because there were 3 sets
of environment variables. The scripts are:
* run-binary.sh: just sets LD_LIBRARY_PATH
* run-jvm-binary.sh: sets LD_LIBRARY_PATH and CLASSPATH
* start-daemon.sh: sets LD_LIBRARY_PATH and CLASSPATH and
kerberos-related environment variables.
The binaries, in almost all cases, work fine without
those tweaks, because libstdc++ and libgcc are picked
up along with libkuduclient.so from the toolchain (they
are in the same directory). I decided to leave good enough
alone here. run-binary.sh and friends can be used in
any remaining edge cases to run binaries.
An alternative to the 3 scripts would be to have an
uber-script that set all the variables, but I felt
that it was better to be specific about what
each binary needed. Cleaning the LD_LIBRARY_PATH
mess up has given me a distaste for scattershot
setting of environment variables. I am open to
revisiting this.
Testing:
* Ran tests on centos 7
* Manually tested that my dev env with
LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu continued
to work (for now). All ubuntu 16.04 and 18.04 dev
envs that were set up with bootstrap_development.sh
will be in this state.
Change-Id: I61c83e6cca6debb87a12135e58ee501244bc9603
Reviewed-on: http://gerrit.cloudera.org:8080/14494
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
This is the main patch for making the the impala-shell cross-compatible with
python 2 and python 3. The goal is wind up with a version of the shell that will
pass python e2e tests irrepsective of the version of python used to launch the
shell, under the assumption that the test framework itself will continue to run
with python 2.7.x for the time being.
Notable changes for reviewers to consider:
- With regard to validating the patch, my assumption is that simply passing
the existing set of e2e shell tests is sufficient to confirm that the shell
is functioning properly. No new tests were added.
- A new pytest command line option was added in conftest.py to enable a user
to specify a path to an alternate impala-shell executable to test. It's
possible to use this to point to an instance of the impala-shell that was
installed as a standalone python package in a separate virtualenv.
Example usage:
USE_THRIFT11_GEN_PY=true impala-py.test --shell_executable=/<path to virtualenv>/bin/impala-shell -sv shell/test_shell_commandline.py
The target virtualenv may be based on either python3 or python2. However,
this has no effect on the version of python used to run the test framework,
which remains tied to python 2.7.x for the foreseeable future.
- The $IMPALA_HOME/bin/impala-shell.sh now sets up the impala-shell python
environment independenty from bin/set-pythonpath.sh. The default version
of thrift is thrift-0.11.0 (See IMPALA-9489).
- The wording of the header changed a bit to include the python version
used to run the shell.
Starting Impala Shell with no authentication using Python 3.7.5
Opened TCP connection to localhost:21000
...
OR
Starting Impala Shell with LDAP-based authentication using Python 2.7.12
Opened TCP connection to localhost:21000
...
- By far, the biggest hassle has been juggling str versus unicode versus
bytes data types. Python 2.x was fairly loose and inconsistent in
how it dealt with strings. As a quick demo of what I mean:
Python 2.7.12 (default, Nov 12 2018, 14:36:49)
[GCC 5.4.0 20160609] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> d = 'like a duck'
>>> d == str(d) == bytes(d) == unicode(d) == d.encode('utf-8') == d.decode('utf-8')
True
...and yet there are weird unexpected gotchas.
>>> d.decode('utf-8') == d.encode('utf-8')
True
>>> d.encode('utf-8') == bytearray(d, 'utf-8')
True
>>> d.decode('utf-8') == bytearray(d, 'utf-8') # fails the eq property?
False
As a result, this was inconsistency was reflected in the way we handled
strings in the impala-shell code, but things still just worked.
In python3, there's a much clearer distinction between strings and bytes, and
as such, much tighter type consistency is expected by standard libs like
subprocess, re, sqlparse, prettytable, etc., which are used throughout the
shell. Even simple calls that worked in python 2.x:
>>> import re
>>> re.findall('foo', b'foobar')
['foo']
...can throw exceptions in python 3.x:
>>> import re
>>> re.findall('foo', b'foobar')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/data0/systest/venvs/py3/lib/python3.7/re.py", line 223, in findall
return _compile(pattern, flags).findall(string)
TypeError: cannot use a string pattern on a bytes-like object
Exceptions like this resulted in a many, if not most shell tests failing
under python 3.
What ultimately seemed like a better approach was to try to weed out as many
existing spurious str.encode() and str.decode() calls as I could, and try to
implement what is has colloquially been called a "unicode sandwich" -- namely,
"bytes on the outside, unicode on the inside, encode/decode at the edges."
The primary spot in the shell where we call decode() now is when sanitising
input...
args = self.sanitise_input(args.decode('utf-8'))
...and also whenever a library like re required it. Similarly, str.encode()
is primarily used where a library like readline or csv requires is.
- PYTHONIOENCODING needs to be set to utf-8 to override the default setting for
python 2. Without this, piping or redirecting stdout results in unicode errors.
- from __future__ import unicode_literals was added throughout
Testing:
To test the changes, I ran the e2e shell tests the way we always do (against
the normal build tarball), and then I set up a python 3 virtual env with the
shell installed as a package, and manually ran the tests against that.
No effort has been made at this point to come up with a way to integrate
testing of the shell in a python3 environment into our automated test
processes.
Change-Id: Idb004d352fe230a890a6b6356496ba76c2fab615
Reviewed-on: http://gerrit.cloudera.org:8080/15524
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Adds impala-shell support to connect to HiveServer2 HTTP endpoint.
Relies on toolchain change at https://gerrit.cloudera.org/#/c/13725/.
Use --protocol='hs2-http' to enable this behavior.
Example usages:
---------------
impala-shell --protocol='hs2-http' (No auth)
impala-shell --protocol='hs2-http' --ldap -u..... (PLAIN auth)
impala-shell --protocol-'hs2-http' --ssl --ca_cert... (TLS)
impala-shell --protocol='hs2-http' --ldap --ssl --ca_cert... (LDAP +
TLS)
Limitations:
-----------
- Does not support Kerberos (-k) due to lack ot SPNEGO support.
Testing:
--------
- Parameterized existing shell tests to support this combination.
- Added shell test coverage for LDAP auth.
Change-Id: I8323950857dfe1c1dfd5377fde79f87bc2ce9534
Reviewed-on: http://gerrit.cloudera.org:8080/13746
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Reviewed-by: Bharath Vissapragada <bharathv@cloudera.com>
Apparently IMPALA_REMOTE_URL is not generally used for remote cluster
tests: only --testing_remote_cluster is reliably set. Fix the
is_remote_cluster() implementation to take into account
REMOTE_DATA_LOAD and --testing_remote_cluster in addition to
IMPALA_REMOTE_URL. Consistently use is_remote_cluster() in
other tests instead of checking the pytest flag directly.
There were a few lifecycle headaches with how
ImpalaTestClusterProperties is used:
* common.environ is imported from conftest, which means that
the top-level code in the file runs *before* pytest
command-line arguments have been registered and parsed.
* ImpalaTestClusterProperties is used by various code,
like build_flavor_timeout(), which runs before pytest
command-line arguments have been parsed.
* ImpalaTestClusterProperties is called from non-pytest
scripts like start-impala-cluster.py, so the command-line
arguments are not available.
I dealt with the above challenges by making a few changes
to do the detection later:
* Lazily initializing a singleton ImpalaTestClusterProperties.
This was not strictly necessary but makes the whole problem
less sensitive to import order and module dependencies.
* Adding cluster_properties fixture to make ImpalaTestClusterProperties
available in tests without additional boilerplate.
* Removing the caching of the local/remote build calculation.
ImpalaTestClusterProperties is instantiated outside of python
tests, but is_remote_cluster() is only called from python tests,
so if we check flags in is_remote_cluster() we'll get the
right results reliably.
As a workaround to unblock remote tests, also assume catalog_v1 if
accessing the web UI fails.
Testing:
Ran core tests against a regular minicluster.
Ran tests against a remote cluster
Change-Id: Ifa6b2a1391f53121d3d7c00c5cf0a57590899ce4
Reviewed-on: http://gerrit.cloudera.org:8080/13386
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
HS2 is added as an option via --protocol=hs2. The user-visible
differences in behaviour are minimal. Beeswax is still the
default and can be explicitly enabled via --protocol=beeswax
but will be deprecated. The default is unchanged because
changing the default could break certain workflows, e.g.
those that explicitly specify the port with -i or deployments
that hit --fe_service_threads for HS2 and somehow rely on
impala-shell not contributing to that limit. For most
workflows the change is transparent and we should change
the default in a major version change.
This support requires Impala-specific extensions to
the HS2 interface, similar to the existing extensions
to Beeswax. Thus the HS2 shell is only
forwards-compatible with newer Impala versions.
I considered trying to gracefully degrade when the
new extensions weren't present, but it didn't seem to be
worth the ongoing testing effort.
Differences between HS2 and Beeswax are abstracted into
ImpalaClient subclasses.
Here are the changes required to make it work:
* Switch to TBinaryProtocolAccelerated to avoid perf
regression. The HS2 protocol requires decoding
more primitive values (because its not a string-per-row),
which was slow with the pure python implementation of
TBinaryProtocol.
* Added bitarray module to efficiently unpack null indicators
* Minimise invasiveness of changes by transposing and stringifying
the columnar results into rows in impala_client.py. The transposition
needs to happen before display anyway.
* Add PingImpalaHS2Service() to get back version string and webserver
address.
* Add CloseImpalaOperation() extension to return DML row counts. This
possibly addresses IMPALA-1789, although we need to confirm that
this is a sufficient solution.
* Add is_closed member to query handles to avoid shell independently
tracking whether the query handle was closed or not.
* Include query status in HS2 log to match beeswax.
* HS2 GetLog() command now includes query status error message for
consistency with beeswax.
* "set"/"set all" uses the client requests options, not the session
default. This captures the effective value of TIMEZONE, which
was previously missing. This also requires test changes where
the tests set non-default values, e.g. for ABORT_ON_ERROR.
* "set all" on the server side returns REMOVED query options - the
shell needs to know these so it can correctly ignore them.
* Clean up self.orig_cmd/self.last_leading comment argument
passing to avoid implicit parameter passing through multiple
function calls.
* Clean up argument handling in shell tests to consistently pass
around lists of arguments instead of strings that are subject
to shell tokenisation rules.
* Consistently close connections in the shell to avoid leaking
HS2 sessions. This is enforced by making ImpalaShell a context
manager and also eliminating all sys.exit() calls that would
bypass the explicit connection closing.
Testing:
* Shell tests can run with both protocols
* Add tests for formatting of all types and NULL values
* Added testing for floating point output formatting, which does
change as a result of switching to server-side vs client-side
formatting.
* Verified that newly-added tests were actually going through HS2
by disabling hs2 on the minicluster and running tests.
* Add checks to test_verify_metrics.py to ensure that no sessions
are left open at the end of tests.
Performance:
Baseline from beeswax shell for large extract is as follows:
$ time impala-shell.sh -B -q 'select * from tpch_parquet.orders' > /dev/null
real 0m6.708s
user 0m5.132s
sys 0m0.204s
After this change it is somewhat slower, but we generally don't consider
bulk extract performance through the shell to be perf-critical:
real 0m7.625s
user 0m6.436s
sys 0m0.256s
Change-Id: I6d5cc83d545aacc659523f29b1d6feed672e2a12
Reviewed-on: http://gerrit.cloudera.org:8080/12884
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Currently, impalarc files can be specified on a per-user basis
(stored in ~/.impalarc), and they aren't created by default. The
Impala shell should pick up /etc/impalarc as well, in addition
to the user-specific configurations.
The intent here is to allow a "global" configuration of the shell
by a system administrator. The default path of the global config
file can be changed by setting the $IMPALA_SHELL_GLOBAL_CONFIG_FILE
environment variable.
Note that the options set in the user config file take precedence
over those in the global config file.
Change-Id: I3a3179b6d9c9e3b2b01d6d3c5847cadb68782816
Reviewed-on: http://gerrit.cloudera.org:8080/13313
Reviewed-by: Bikramjeet Vig <bikramjeet.vig@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
shell/make_shell_tarball.sh builds a tarball with all the
shell dependencies bundled. We should test the contents of
that tarball in the shell tests instead of using infra/python/env
and the libraries bundled there.
This tarball is one of the default targets (e.g. run by buildall.sh) so
this should not affect any typical development workflows.
Note that this means the shell tests now requires the shell tarball to
be built locally, which doesn't necessarily happen for remote cluster
tests, so we preserve the old behaviour in that case.
Testing:
Ran core tests on CentOS 6 and CentOS 7.
Change-Id: I581363639b279a9c2ff1fd982bdb140260b24baa
Reviewed-on: http://gerrit.cloudera.org:8080/13267
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
This sets up the tests to be extensible to test shell
in both beeswax and HS2 modes.
Testing:
* Add test dimension containing only beeswax in preparation
for HS2 dimension.
* Factor out hardcoded ports.
* Add tests for formatting of all types and NULL values.
* Merge date shell test into general type tests.
* Added testing for floating point output formatting, which does
change as a result of switching to server-side vs client-side
formatting.
* Use unique_database for tests that create tables.
Change-Id: Ibe5ab7f4817e690b7d3be08d71f8f14364b84412
Reviewed-on: http://gerrit.cloudera.org:8080/13083
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
test_ssl has a logic that waits for the number of in-flight queries to
be 1. However, the logic for wait_for_num_in_flight_queries(1) only
waits for the condition to be true for a period of time and does not
throw an exception when the time has elapsed and the condition is not
met. In other words, the logic in test_ssl that loops while the number
of in-flight queries is 1 never gets executed. I was able to simulate
this issue by making Impala shell start much longer.
Prior to this patch, in the event that Impala shell took much longer to
start, the test started sending the commands to Impala shell even when
Impala shell was not ready to receive commands. The patch fixes the
issue by waiting until Impala shell is connected. The patch also adds
assert in other places that calls wait_for_num_in_flight_queries and
updates the default behavior for Impala shell to wait until it is
connected.
Testing:
- Ran core and exhaustive tests several times on CentOS 6 without any
issue
Change-Id: I9805269d8b806aecf5d744c219967649a041d49f
Reviewed-on: http://gerrit.cloudera.org:8080/12047
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
test_cancellation runs a shell process, executes a query, sleeps,
sends a sigint to the process, and then checks that the query is
cancelled. If the sigint is sent prior to the shell installing its
signal handler, the test can fail with a KeyboardInterrupt.
This patch removes the reliance on the sleep being long enough by
actually reading the output of the shell and only cancelling the
query once the shell shows that it has started running.
Testing:
- Ran test_cancellation in a loop.
Change-Id: I65302ffb838d5185f77853bc2e53296f3a701d93
Reviewed-on: http://gerrit.cloudera.org:8080/11255
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Reviewed-by: Thomas Marshall <thomasmarshall@cmu.edu>
This adds an IMPALA_HISTFILE environment variable (and --history_file
argument) to the shell which overrides the default location of
~/.impalahistory for the shell history. The shell tests now override
this variable to /dev/null so they don't store history. The tests that
need history use a pytest fixture to use a temporary file for their
history. This allows so that they can run in parallel without stomping
on each other's history.
This also fixes a couple flaky test which were previously missing the
"execute_serially" annotation -- that annotation is no longer needed
after this fix.
A couple of the tests still need to be executed serially because they
look at metrics such as the number of executed or running queries, and
those metrics are unstable if other tests run in parallel.
I tested this by running:
./bin/impala-py.test tests/shell/test_shell_interactive.py \
-m 'not execute_serially' \
-n 80 \
--random
... several times in a row on an 88-core box. Prior to the change,
several would fail each time. Now they pass.
Change-Id: I1da5739276e63a50590dfcb2b050703f8e35fec7
Reviewed-on: http://gerrit.cloudera.org:8080/11045
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Reviewed-by: Todd Lipcon <todd@apache.org>
Jenkins jobs occasionally hang on test_query_cancellation_during_fetch.
There was a workaround proposal submitted under this Jira ID, however,
apparently jobs still hang on this test randomly. Reverting the
workaround and skipping the test until further fix proposal provided.
This reverts commit 7810d1f9a2.
Change-Id: I51acee49b5a17c4852410b7568fd1d092b114a6d
Reviewed-on: http://gerrit.cloudera.org:8080/8972
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
Apparently test_query_cancellation_during_fetch hangs occasionally
in Jenkins builds. The Impala debug page shows the query being
cancelled, however, on the host the ImpalaShell process related to
that query is still running.
Since I had no luck in reproducing the issue locally I only have a
theory what might be going on here: The query is cancelled
successfully on Impala backend and when the test tries to get the
stdout and stderr from the ImpalaShell it gets stuck. It might be
the case that ImpalaShell process fetching the query results holds
the stdout. According to the documentation of subprocess.communicate()
it may cause issues to fetch data when the data size is large or
unlimited, that we can consider to be the case here.
As a workaround there is a new optional parameter to
util.ImpalaShell to omit the stdout because this test wouldn't use
it anyway and we get rid of fetching the large result from
ImpalaShell.
Change-Id: I082c83b91b6d0c527de92c7992f0dc9d1b290433
Reviewed-on: http://gerrit.cloudera.org:8080/8852
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
The ImpalaShell didn't issue the 'USE <current-db>' command after
reconnecting to the Impala daemon. Therefore the client session
used the default DB after reconnection, not the previously selected DB.
Setting the current DB is done by the _validate_database method.
Before this commit it appended the "use <db>" command to the
command queue of the Cmd class. But, at this point we might already
have commands in the command queue that will run before the
"use <db>" command. In case of reconnection, we want to invoke
the USE command right away.
Also, the command processed by the precmd() method can entirely skip
the command queue, therefore it is not enough to insert the USE
command to the front of the command queue. We need to issue the
USE command with the onecmd() method to execute it immediately.
I extended the _validate_database method with an "immediately" flag.
If this flag is true, _validate_database will use the onecmd() method.
Otherwise, it will append the USE command to the command queue to
maintain the previous behaviour.
I added a new automated test suite named test_shell_interactive_reconnect.py
to the "custom cluster" tests. It sets the default database, and after
reconnection it checks if the shell set it again automatically.
One test case checks if the shell set the DB after manually reconnecting
to the impala daemon by issuing the CONNECT command.
The other test case checks if the shell set the DB after automatic
reconnection due to cluster restart.
I needed to backup the impala shell history file because I didn't
want to pollute it by the test cases (just like the way it is done in
tests/shell/test_shell_interactive.py). I created utility functions for
this in tests/shell/util.py and now test_shell_interactive.py and
the newly created test suite are using these utility functions.
Change-Id: I40dfa00ba0314d356fe8617446f516505c925e5e
Reviewed-on: http://gerrit.cloudera.org:8080/8368
Reviewed-by: Tim Armstrong <tarmstrong@cloudera.com>
Tested-by: Impala Public Jenkins
The plan-root fragment instance that runs on the coordinator should be
handled like all others: started via RPC and run asynchronously. Without
this, the fragment requires special-case code throughout the
coordinator, and does not show up in system metrics etc.
This patch adds a new sink type, PlanRootSink, to the root fragment
instance so that the coordinator can pull row batches that are pushed by
the root instance. The coordinator signals completion to the fragment
instance via closing the consumer side of the sink, whereupon the
instance is free to complete.
Since the root instance now runs asynchronously wrt to the coordinator,
we add several coordination methods to allow the coordinator to wait for
a point in the instance's execution to be hit - e.g. to wait until the
instance has been opened.
Done in this patch:
* Add PlanRootSink
* Add coordination to PFE to allow coordinator to observe lifecycle
* Make FragmentMgr a singleton
* Removed dead code from Coordinator::Wait() and elsewhere.
* Moved result output exprs out of QES and into PlanRootSink.
* Remove special-case limit-based teardown of coordinator fragment, and
supporting functions in PlanFragmentExecutor.
* Simplified lifecycle of PlanFragmentExecutor by separating Open() into
Open() and Exec(), the latter of which drives the sink by reading
rows from the plan tree.
* Add child profile to PlanFragmentExecutor to measure time spent in
each lifecycle phase.
* Removed dependency between InitExecProfiles() and starting root
fragment.
* Removed mostly dead-code handling of LIMIT 0 queries.
* Ensured that SET returns a result set in all cases.
* Fix test_get_log() HS2 test. Errors are only guaranteed to be visible
after fetch calls return EOS, but test was assuming this would happen
after first fetch.
Change-Id: Ibb0064ec2f085fa3a5598ea80894fb489a01e4df
Reviewed-on: http://gerrit.cloudera.org:8080/4402
Tested-by: Internal Jenkins
Reviewed-by: Henry Robinson <henry@cloudera.com>
For files that have a Cloudera copyright (and no other copyright
notice), make changes to follow the ASF source file header policy here:
http://www.apache.org/legal/src-headers.html#headers
Specifically:
1) Remove the Cloudera copyright.
2) Modify NOTICE.txt according to
http://www.apache.org/legal/src-headers.html#notice
to follow that format and add a line for Cloudera.
3) Replace or add the existing ASF license text with the one given
on the website.
Much of this change was automatically generated via:
git grep -li 'Copyright.*Cloudera' > modified_files.txt
cat modified_files.txt | xargs perl -n -i -e 'print unless m#Copyright.*Cloudera#i;'
cat modified_files_txt | xargs fix_apache_license.py [1]
Some manual fixups were performed following those steps, especially when
license text was completely missing from the file.
[1] https://gist.github.com/anonymous/ff71292094362fc5c594 with minor
modification to ORIG_LICENSE to match Impala's license text.
Change-Id: I2e0bd8420945b953e1b806041bea4d72a3943d86
Reviewed-on: http://gerrit.cloudera.org:8080/3779
Reviewed-by: Dan Hecht <dhecht@cloudera.com>
Tested-by: Internal Jenkins
The impala-shell could not accept wildcard or SAN certificates
previously as the thrift library it depended on did not support them.
This patch subclasses TSSLSocket and adds the logic to take care of
the above mentioned cases by introducing the new
TSSLSocketWithWildcardSAN class.
The certificate matching logic is based on the python-ssl source code.
Added custom cluster tests to test both wildcard matching and SAN
matching.
Added be/src/testutil/certificates-info.txt which contains all the
information about the certificates which are added for the tests.
This has been tested with Python2.4 and Python2.6.
Change-Id: I75e37012eeeb0bcf87a5edf875f0ff915daf8b89
Reviewed-on: http://gerrit.cloudera.org:8080/3765
Reviewed-by: Sailesh Mukil <sailesh@cloudera.com>
Tested-by: Internal Jenkins
To cancel a query, the shell will create a separate connection inside
it's SIGINT handler, and send the cancellation RPC. However this
connection did not start a secure connection if it needed to, meaning
that the cancellation attempt would just hang.
A workaround is to kill the shell process, which I expect is what users
have been doing with this bug which has been around since 2014.
Testing:
I added a custom cluster test that starts Impala with SSL
enabled, and wrote two tests - one just to check SSL connectivity, and
the other to mimic the existing test_cancellation which sends SIGINT to
the shell process. In doing so I refactored the shell testing code a bit
so that all tests use a single ImpalaShell object, rather than rolling
their own Popen() based approaches when they needed to do something
unusual, like cancel a query.
In the cancellation test on my machine, SIGINT can take a few tries to
be effective. I'm not sure if this is a timing thing - perhaps the
Python interpreter doesn't correctly pass signals through to a handler
if it's in a blocking call, for example. The test reliably passes within
~5 tries on my machine, so the test tries 30 times, once per second.
Change-Id: If99085e75708d92a08dbecf0131a2234fedad33a
Reviewed-on: http://gerrit.cloudera.org:8080/3302
Reviewed-by: Henry Robinson <henry@cloudera.com>
Tested-by: Henry Robinson <henry@cloudera.com>
