I have been consistently receiving the same error while trying to create a connection between R and Snowflake using an ODBC driver. The error that I'm receiving says:
'''Error during wrapup: nanodbc/nanodbc.cpp:1021: 00000: [unixODBC][Driver Manager]Can't open lib 'Snowflake' : file not found
Error: no more error handlers available (recursive errors?); invoking 'abort' restart'''
The instructions given by Snowflake to connect a driver are rather cut and clear - but provides little insight on what to do with errors. Can anyone lead me as to where to go given this error?
This smells like a configuration issue. Your driver manager, unixODBC, isn't able to locate the Snowflake driver. (The Snowflake documentation says on MacOS to use iODBC. Don't. It won't work with the odbc R package.)
Check that unixODBC is findable. Run odbcinst -j in a terminal. If it works, you will know that you have unixODBC properly installed, and it will give you the paths to your various configuration files.
On to checking configuration. This is the documentation for configuring on Linux using unixODBC. If you are using MacOS the same general instructions apply, but the file extensions will change from .so to .dylib . Since it's saying it can't find the file, I'm thinking that using full paths might resolve this for you. It's also possible that there's some issue with how you are specifying the driver.
Also, it looks like it's searching for a file named 'Snowflake'. I'm thinking you've got Driver=Snowflake somewhere in one of your config files. Best change that to Driver=<path>/<to>/<driver>/libSnowflake.dylib (or .so if you're on Linux). Do this in all the places where you have Driver=Snowflake.
Related
I was running across this piece of documentation: https://cmake.org/cmake/help/v3.13/manual/cmake-server.7.html
But as it turns out, the description given on how to start the server, is not the same as shown in the documentation. When executing cmake -E server --debug, I am getting the following message:
CMake Error: No protocol versions defined. Maybe you need --experimental?
When adding the suggested flag --experimental, the server will launch, reading from standard input and responding to standard output, as expected. However, all the examples shown on the documentation do not suggest the use of this flag. Plus, I have version 3.13.4 installed as well.
What is the proper way of starting the CMake server, as in how would I specify the protocol version, and why am I asked to use --experimental?
My main goal is to find out if I can extract values like targets and their associated information (like CFLAGS) off CMake. Using this server command would certainly let me do this by just writing a small piece of NodeJS code to interact with it.
I am trying to connect to an Vertica DB from R using "RODBC" package. Also, the machine I am using is an remote server which doesn't have direct internet access so I basically "transfer" all source files from my local to the remote server to build the system. So, in order to give you an clear context, I am listing all my steps in attending of installing "RODBC" package below:
Step1 - I downloaded the RODBC_1.3-13.tar.gz source file for RODBC and then tried to directly install it with "R CMD INSTALL". However, I encountered error as "ODBC headers sql.h and sqlext.h not found".
Step2 - After a few researches, I found that the installation of "unixodbc-dev" would potentially solve this issue. Therefore, I downloaded all needed dependencies for "unixodbc-dev" and transferred them to the server. As you can see the list:
Therefore, I also successfully installed "unixodbc-dev":
However, another error message appears when I tried to re-install the "RODBC" using "sudo R CMD INSTALL /home/mli/RODBC_1.3-13.tar.gz" in which it returns error "no ODBC driver manager found":
As the message indicates, the installation program can't locate my ODBC driver manager. So, I downloaded "vertica-client-7.2.3-0.x86_64.tar.gz" and unzipped it on the server:
So, now my question is how can I customize the "R CMD INSTALL" command say, using some parameter handles to direct the installation program to locate the driver manager? Or am I trying this in a right direction? Please let me know. Any help would be really appreciated!!! :)
ADDITION:
I have also tried it with JDBC in which the I successfully loaded the "RJDBC" package in R and used the JDBC driver from vertica-client-7.2.3-0.x86_64.tar.gz. Also, I have already had "rJava" installed. However, I have still got an error when I tried to make the connection. I am listing my result below:
I successfully installed the "RJDBC" with "$R CMD INSTALL RJDBC_0.2-5.tar.gz --library=/usr/local/lib/R/site-library/" and then I tried the following scripts in R. All the lines are successfully executed except on the line 16:
Based on the error message, I assumed the version of the JDBC driver that I was using is too new for the Vertica server. So, I was trying to use an older version JDBC driver instead, like the "vertica-jdk5-6.1.0-0.jar" which I have downloaded from this link:http://www.java2s.com/Code/Jar/v/Downloadverticajdk56100jar.htm
So, I moved the file "vertica-jdk5-6.1.0-0.jar" to my home directory on the server and then changed the JDBC driver path in the R script:
As you can see, it still returns error "FATAL: Unsupported frontend protocol 3.6: server supports 3.0 to 3.5". Am I doing it right? Or is there an issue with the new driver that I downloaded? How can make it works? Please, any help will be really appreciated! Thanks!!!
A few things:
First, just do sudo apt-get install r-cran-rodbc. The package was created (by yours truly) in no small part because dealing with unixODBC or iODBC is not fun. But even once you have that, you still need the ODBC driver for Linux from Vertica. And that part is filly.
Second, I just did something similar the other day but just used JDBC, which worked. You do of course need sudo apt-get install r-cran-rjava which has its own can of worms (but I already mentioned Java...) Still, maybe try that instead?
Third, you can cheat and just use psql pointed to the Vertica port (usually one above the PostgreSQL port).
I have installed PyDev successfully, I have python 3.4.0 installed successfully and the folder containing python.exe is in my path and I can run it from the command line.
However, when I go to window\preferences\pydev\interpreters\python interpreter and select 'quick auto-config' to configure it using normal python I get an error saying 'auto configurator could not find a valid interpreter... Note the system environment variables used for Jython are PATH and PYTHONHOME.'
Unless this is a bug in the text I am not selecting Jython (I get a different error if I select IronPython, btw, so this is not a generic error message).
The trace is as follows:
Errors getting info on discovered interpreter(s).
See error log for details.
java.lang.RuntimeException: java.io.IOException: Cannot run program "python": CreateProcess error=2, The system cannot find the file specified
As I say, python.exe is in my path.
You have to install the actual python package separately from here https://www.python.org/download
Close LiClipse/Eclipse => Download Python Here: https://www.python.org/download
=> reopen LiClipse => Auto Config for the interpreter Will WORK!
I found the answer, but I'll leave this up in case anyone else gets it. It's a pretty stupid auto config option if you ask me ;)
In the higher level window\preferences\pydev\interpreters\ preference you need to untick jython and ironpython.
No idea why given I selected the Python sub-preference specifically.
As mention in Orafce Install.orafunc:
..install Orafce functions in the database, either run the orafce.sql script using the pgAdmin SQL tool..
I tried running the orafce--3.0.sql in pgAdmin sql editor. This give me error
ERROR: could not access file "MODULE_PATHNAME": No such file or directory.
What do you mean by module path?
Installed program:
strawberry perl with DBD::Oracle
postgresql 9.3
pgAdmin III
Not fully installed:
ora2pg
I tried installing ora2pg...with a problem.
H:\PostgreSQL\ora2pg-12.1>perl makefile.pl
Unparsable version '' for prerequisite DBD::Oracle at makefile.pl line 553
Generating a dmake-style Makefile
Writing Makefile for Ora2Pg
Writing MYMETA.yml and MYMETA.json
Done...
H:\PostgreSQL\ora2pg-12.1>dmake && dmake install
"Installing default configuration file (ora2pg_dist.conf) to C:\ora2pg"
Appending installation info to C:\strawberry\perl\lib/perllocal.pod
dmake: Warning: -- Target [install] was made but the time stamp has not been up
dated.
Suggested Solution:
I downloaded a copy of orafce from okbob github
Unzip the file to folder D:/Postgresql/orafce-master
I copy only the following files
orafce--unpackaged--3.0.6.sql
orafce--3.0.6.sql
orafce.control
to folder C:\Program Files\PostgreSQL\9.3\share\extension
Then I try running this command in pgAdmin III sql tools.
CREATE EXTENSION orafce;
I received this Warning and Error.
[WARNING ] CREATE EXTENSION orafce
ERROR: syntax error in file "C:/Program Files/PostgreSQL/9.3/share/extension/orafce.control" line 1, near end of line
I checked orafce.control content. It has this config.
# intarray extension
comment = 'Functions and operators that emulate a subset of functions and packages from the Oracle RDBMS'
default_version = '3.0.6'
module_pathname = '$libdir/orafunc'
relocatable = false`
I can't pass to this wall. What seems the problem?
So, you are working with source raw files. You should to compile these files first - and later you can use it. It is relative simply on Unix like platforms, where C compiler is usually available, and pretty hard on MS Windows, where you have to install C compiler first.
I afraid so we lost pgFoundry archive, where was orafce precompiled and packed.
Almost all Linux distributions support orafce directly - and you can install it without compilation from repositories.
see http://wiki.postgresql.org/wiki/Building_and_Installing_PostgreSQL_Extension_Modules
$libdir is symbol, that is used for PostgreSQL extensions directory. It can be different for any platform - and it is replaced inside compilation stage by actual value. MODULE_PATHNAME has similar meaning. In compilation stage is replaced by valid actual path to library with compiled code.
I am sorry - we don't provide a compiled files - mainly due high risk for MS Windows. We have no forces, and tools to maintain all Win safely. In this moment, you can:
try to contact someone who use orafce for windows for backup of orafce installers
try to compile this extension by self (Microsoft Visual Studio Express edition is free and downloadable on internet).
other possibility is migrate database server to Linux - almost all database maintenance and usage is more simply and more robust there (due missing viruses, antiviruses and less resource requests). The Linux is primary platform for Oracle too.
some tutorials:
http://blog.2ndquadrant.com/compiling-postgresql-extensions-visual-studio-windows/
http://www.scribd.com/doc/40725510/Build-PostgreSQL-C-Functions-on-Windows
I'm trying to give pg_bulkload a try.
When I try to use the postgresql executable it provides, I get the following error:
/usr/local/src/pg_bulkload-3.0.1/bin> ./postgresql start -D /pg_data
server starting
/usr/local/src/pg_bulkload-3.0.1/bin>
FATAL: unrecognized configuration parameter "wal_level"
Google turned up an exact match for this error when someone was using a 9.0 version of psql to run a script on an instance of Postgres 8.4. I don't see how that could be related to my case--I have two versions of Postgres, but I'm sure I'm pointing at the right directory... any thoughts are very welcome.
As far as I can tell from the docs, PostgreSQL 9.x supports a configuration parameter named "wal_level", but version 8.4 does not. The postgresql.conf file for my 9.0.something server has that parameter; the one for my 8.4 server does not.
PostgreSQL 9.x server configuration
PostgreSQL 8.4 server configuration
Your error message suggests you're running version 8.4, but it's reading the configuration file for a 9.x server. Check your postgresql.conf and installation process. I'm thinking pg_bulkload might have "helped" you in ways you didn't anticipate.
I think that it can be a bit tricky to install pg_bulkload to the right place if you have more than one version of PostgreSQL installed on your machine. My first problem was that pg_bulkload (version 3.1.6) could not find pg_port library. I copied the library libpgport.a (a static library) to /usr/local/lib where it was found, but this approach is not recommended, because this is only a quick fix that doesn't work at the end. So, very soon there was another problem: "undefined reference to `pstrdup'". I reckon that in pg_bulkload there should be a possibility of pointing out where PostgreSQL is installed. Well, I changed Makefile of pg_bulkload in pg_bulkload-3.1.6/bin, namely line with PG_LIBS: PG_LIBS = $(libpq) -L/current location of your PostgreSQL/PostgreSQL/pgsql/lib -lpgport -lpgcommon. -lpgport has to be added before -lpgcommon. Last but not least, to compile and install pg_bulkload you shoud modify your PATH: PATH=/current location of your PostgreSQL/PostgreSQL/pgsql/bin:$PATH make USE_PGXS=1 [install]; This makes sure that your pg_bulkload will be added to the correct version of PostgreSQL (in my case 9.3). Enjoy!