+1 (non-binding)
Verified hashes, built and ran all tests, built docker and ran all integration
tests. Ran a simple test flow.
I did run into one minor issue, which is if virtualenv is configured with
python3 by default then creating the virtual environment for the pytest
integration tests
Joe,
I figured it out. At least on my box, there appears to be contention between
two unit tests that are using the same folder to test the
MinimalLockingWriteAheadLog. Folder name:
target/testRecoverFileThatHasTrailingNULBytes.
Unit tests:
testRecoverFileThatHasTrailingNULBytesNoTruncation
Ok. You might want to check if there is already a JIRA for that. The
tests on Windows are notoriously bad. Many conditional ignores out
there for Windows runs of tests. That said, this may well be an
actual windows problem and the test is possibly flagging it.
Thanks
On Wed, Nov 22, 2017 at
Joe,
This did resolve the RAT issue, but the unit test I mentioned still fails with
the same error.
--Peter
-Original Message-
From: Joe Witt [mailto:joe.w...@gmail.com]
Sent: Thursday, November 23, 2017 10:22
To: dev@nifi.apache.org
Subject: [EXT] Re: Contrib Check Build - RAT and
Peter
I dont' believe RAT works in parallel builds (-T2 for example).
If I use RAT it is during a full clean build and activated via the
'contrib-check' profile. 'mvn clean install -Pcontrib-check'
Thanks
On Wed, Nov 22, 2017 at 9:19 PM, Peter Wicks (pwicks) wrote:
> I'm
I'm trying to successfully run Contrib Check on my dev box.
Windows 10 x64
jdk1.8.0_91
MVN 3.3.9
I'm using IntelliJ, so my Run looks like:
Command line: -T2 -Drat.skip=true clean install
Profile: contrib-check
If I don't disable RAT (-Drat.skip=true) then I get too many RAT failures for
the
+1, binding
Built, tested, and created Docker container on OS X 10.12 and Centos 7.3
Ran a few flows and verified expected functionality on both systems.
Thanks for getting this RC together!
On Wed, Nov 22, 2017 at 10:49 AM, Jeremy Dyer wrote:
> Your right let’s not put out
Thanks, I narrowed it down to not having libbz2 for that failure, liblzma
will save me another round of building.
On Wed, Nov 22, 2017 at 5:08 PM, Marc wrote:
> Tony,
> I hit send before adding the full commands.
>
> sudo apt-get install libbz2-dev liblzma-dev .
>
>
Tony,
I hit send before adding the full commands.
sudo apt-get install libbz2-dev liblzma-dev .
That would make the command in the helper guide:
sudo apt-get cmake gcc g++ libcurl4-openssl-dev uuid-dev uuid
libboost-all-dev libssl-dev doxygen libpython3-dev libbz2-dev liblzma-dev .
Tony,
Try adding libbz2-dev liblzma-dev to your distro if you haven't already.
I don't think I correctly copied the command for apt-get. My apologies.
On Wed, Nov 22, 2017 at 4:26 PM, Tony Kurc wrote:
> I'm having a bit of trouble with running make test - still diagnosing,
I'm having a bit of trouble with running make test - still diagnosing, but
it is a segfault. (building on an ubuntu 16.04 docker image).
---
CompressFileBZip
Dear Devs,
I have recently implemented two-way SSL authentication for a NiFi cluster. The
tls toolkit (nigh-toolkit) made that process fairly easy for both the cluster
and for generating client certificates. I now have a situation where I need to
be able to revoke access for a client
Matthew,
It's awesome that you've got these kinds of tests running! Thanks for shooting
a note
to the dev list. After a quick glance at the code, I can understand why the
second case
was throwing an Exception. I'm not sure why the other case was occurring. But I
am happy
to look into it.
I've
Hello,
In updating our Nifi setup to 1.4.0, one of our regression tests started to
fail.
On investigation I ran into an issue with FlowFile state that has confused
me.
The following processor takes an input flow file, writes "NEW_DATA" to it,
the clones it.
It then outputs the original flowFile
AFAIK this is not currently possible, if you want to use a file then
you must restart NiFi. Alternatively as of NiFi 1.4.0, you can set
variables on a process group (including the top-level "root" group),
you can change those without restarting NiFi. I answered a similar
question on StackOverflow
Hi,
i have declared my custom properties file path in nifi.properties to
variable 'nifi.variable.registry.properties=custom.properties'
.i want my processors to pickup the new value of the property when the
chnages are made to a property.I have loaded the updated properties in the
I have an UpdateAttribute processor with advanced rules that will store a
counter for each of 239 attributes in the local state. This appears to
work just fine, as I see 956 total state items in the View State view (I
have 4 nodes in my cluster). However, I can't seem to retrieve more than
500
Hi,
i have declared my custom properties file path in nifi.properties to
variable 'nifi.variable.registry.properties=custom.properties'
.i want my processors to pickup the new value of the property when the
chnages are made to a property.I have loaded the updated properties in the
Your right let’s not put out another release for this since it can be simply
fixed by using the flags you provided. I went through the build again and
validated the runtime. Everything looks good now so I’m changing my vote to a
+1
- Jeremy
> On Nov 22, 2017, at 10:31 AM, Marc
Jeremy,
Thanks for your vote.
Your version of GCC will likely cause this warnings due to spec
additions, and since RocksDB fails on any warning your build failed as well.
Please try the following before running make: *cmake -DPORTABLE=ON
-DFAIL_ON_WARNINGS= ..*
I am not in favor of
-1 Marc I'm having trouble getting this to build using Ubuntu 17.10 with
GCC version "gcc (Ubuntu 7.2.0-8ubuntu3) 7.2.0" seems to be an issue with
building RocksDB with this version of GCC. Looks like there is an update of
RocksDB where this would work however. What do you think?
On Tue, Nov 21,
21 matches
Mail list logo