Installation issue: py-matplotlib #11396
Comments
|
I'll try to reproduce this tomorrow. It looks like you're on a RHEL system (RHEL7?) - is that right? |
|
RHel 6.9 Thank you. |
|
I'm having trouble replicating this, but I have only been able to try RHEL 7 so far. A few questions:
|
|
Thanks for the reply I don't have any external packages recorded in packages.yaml Python 3.7.3 successfully installed but fails for matplotlib |
|
Do you need any more info? Thanks. |
|
You need to use #7926 |
|
...then you need to turn on the |
|
Thanks for this but when creating an env to the bottom I get File "/shared/spack/bin/spack", line 48, in I noticed in your example the file is called discover.yaml |
My example uses Spack Environments; looks like you're using that too. Here is the main environment file, which links to the config file: It looks to me like you're tackling environments and Python3 all at the same time. It might be easier to do one at a time. You should be able to put this in your Then you should be able to successfully do Even without messing with YAML files, you should be able to do @tgamblin We REALLY need to get serious about supporting Python3 without jumping through hoops. |
|
Thank you again. But when I added that and ran spack spec py-matplotlib it gives me ==> Error: An unsatisfiable version constraint has been detected for spec: while trying to concretize the partial spec: py-matplotlib requires python version 3.7.3:, but spec asked for 2.7.15 |
|
@liamrpowell I just did the following (on my Mac; but Linux should be about the same). Use for Then: Result is: If my for of Spack works for you, feel free to use it. Sorry I don't have more time to make this work on the latest Spack |
|
@citibeth I'm confused: I don't see anything in which mentions the error and furthermore I was able to install I am more inclined to suspect this is an issue with |
|
Sorry, I don't know what the problem is. It works for me using the repo I shared above, on CentOS7. |
|
@liamrpowell could you attach the following files (running with a "fresh" Spack configuration, after reproducing the installation error):
|
|
I will next week. Had other work on our supercomputer that took precedence
first.
Just to be clear, you want me to get rid of the current speck install and
start from scratch?
On May 15, 2019 at 3:49:44 PM, Scheibel Peter (notifications@github.com) wrote:
@liamrpowell <https://github.com/liamrpowell> could you attach the
following files (running with a "fresh" Spack configuration, after
reproducing the installation error):
- The spack-build.log that appears in the staging directory (you can spack
cd py-matplotlib ^python@3: to CWD to the directory with that file after
a failed build).
- The full output of spack spec py-matplotlib ^python@3:?
- Looking at the py-matplotlib documentation (
https://matplotlib.org/3.0.2/users/installing.html) there appears to be
a separate (not Spack-maintained) log - could you include that as well?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#11396?email_source=notifications&email_token=ABS336KBYJXWI62ROYHIEELPVRSNRA5CNFSM4HLTVACKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODVPYDNQ#issuecomment-492798390>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABS336I6KQ4ZEM2TJ52ADKDPVRSNRANCNFSM4HLTVACA>
.
|
|
Fresh install. did spack install py-matplotlib^python@3: ==> Error: An unsatisfiable version constraint has been detected for spec: while trying to concretize the partial spec: py-matplotlib requires python version 3.5:, but spec asked for 2.7.15 spack spec py-matplotlib ^python@3: Input specpy-matplotlib Concretizedpy-matplotlib@3.0.0%gcc@4.8.2 |
|
Something in your concretization requires Python 2.7. Could it be one of these packages??? I would check Can you start out with the fork I recommend above, and put in the right |
|
git@github.com: Permission denied (publickey). On your fork you provided. |
|
You need:
git clone https://github.com/citibeth/spack.git
…On Mon, May 20, 2019 at 11:59 AM liamrpowell ***@***.***> wrote:
***@***.***: Permission denied (publickey).
fatal: Could not read from remote repository.
On your fork you provided.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#11396?email_source=notifications&email_token=AAOVY523JSQA34KEX6WHFZLPWLDG7A5CNFSM4HLTVACKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODVZJKNQ#issuecomment-494048566>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAOVY56WWBPPNLYISXXL5CTPWLDG7ANCNFSM4HLTVACA>
.
|
|
I cloned the main git repo (on a Mac), got rid of my `~/.spack` directory
and did `spack spec py-matplotlib ^python@3:`. Worked for me.
|
|
One confusing thing is there are apparently two problems mentioned in this
thread: one with concretization and one with building. Previously we saw
some build errors. Right now we seem to be looking at concretization
errors.
|
|
Ok i cloned your git. Then added the python to spack/defaults/packages.yaml then ran spack spec py-matplotlib ^python@3: and get Error: /shared/spack/etc/spack/defaults/packages.yaml:16: Additional properties are not allowed ('python', 'all' were unexpected) |
|
All the lines except the first need to be indented an additional indent. |
|
all seemed to be going well until ran into this 2 errors found in build log:
See build log for details: |
|
I did try set FORCE_UNSAFE_CONFIGURE=1 btw |
|
Wait a minute... are you running Spack as root or with |
No, please don't do this. |
|
Ok let me try something else. |
|
running it as sudo under my user account (not root) |
|
Running as `sudo` under a user account is functionally the same as running
as root.
This is your obvious problem: Spack is not designed to run as root. There
is no point in doing so, and it could also be a security problem. What's
happening here... Spack doesn't check / care whether you run as root
(although you shouldn't). But the `tar` upstream package notices Spack
(running as root) is running `./configure` as root, which is (known to be)
a security problem. So it's correctly refusing to run. You are receiving
an error message from the `tar` build, not from Spack proper.
In any case, you need to start over, running Spack in your regular user
account NOT as root.
|
|
Ok I thought you had to run spack installl using sudo. My mistake. |
|
Yes... because some files in |
|
What should the permissions on the spack folder be? This is for multiple users as it is in the /shared folder |
|
On Mon, May 20, 2019 at 2:47 PM liamrpowell ***@***.***> wrote:
What should the permissions on the spack folder be? This is for multiple
users as it is in the /shared folder
It should be readable by the people you intend to use it; and you should
set up `umask` accordingly. If it were me, and I had root permission, I
would set up a new user for the purpose of running Spack (maybe call the
user `spack`). Then I would set `umask` and groups properly in that user's
`.bashrc` so that all my intended users can read the result. You shouldn't
be doing `chmod` or `chgrp` on the Spack-installed packages, that could
cause problems.
BTW... @baberlevi <https://github.com/baberlevi> reports that Spack runs
OK as root, as long as you do the `FORCE_UNSAFE_CONFIGURE` flag. I still
don't recommend it because it's a massive violation of the principle of
least privilege:
https://en.wikipedia.org/wiki/Principle_of_least_privilege
… |
|
OK got further. but ran into this error for Matplotlib. |
|
Please use triple-backquotes to quote your console output. Fixing up now so I can read it... |
|
Problems I see here; hopefully someone else has a good idea on how to solve them:
Many / most useful Python stacks involve compiling C/C++ code (for example,
Most reliable would be to use a single GCC that includes all the languages you need. You can get this with "Mixed" compilers have been successful when combining Clang with Gfortran on macOS. But I'm worried about combining two differnt versions of GCC; more chance (in my mind) that their libraries or expacations might clash. |
|
I will do the "intentionally trying to build py-matplotlib with the system Python (don't)" When i tried to do is that forcing it to use the local Python install instead of the Python within spack? |
|
When i tried to do spack spec py-matplotlib ***@***.***:
is that forcing it to use the local Python install instead of the Python
within spack?
No, it is telling Spack which (Spack-built) version of Python to use. You
tell Spack to use system versions of packages by putting stuff in
`packages.yaml`.
I don't really know how `/usr/local/lib` is getting in your build paths.
But that's the key to resolving your problem.
|
|
Ok here we go again. trying to install GCC. |
|
Something is strange with your system. Does this help?
https://stackoverflow.com/questions/33734143/gcc-unable-to-find-shared-library-libisl-so
…On Tue, May 21, 2019 at 11:07 AM liamrpowell ***@***.***> wrote:
Ok here we go again.
trying to install GCC.
``` Error:` ProcessError: Command exited with status 2:
'make' '-j6'
6 errors found in build log:
9722 /tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0/spack-build/
./prev-gcc/xg++ -B/tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-
8.2.0/spack-build/./prev-gcc/ -B/shared/spack/opt/spack/linux-rhel
6-x86_64/gcc-4.8.2/gcc-8.2.0-arepu3ta2xtlboacnuxk3su2h3oqc3lr/x86_
64-pc-linux-gnu/bin/ -nostdinc++ -B/tmp/wpowell/spack-stage/spack-
stage-u_7lic/gcc-8.2.0/spack-build/prev-x86_64-pc-linux-gnu/libstd
c++-v3/src/.libs -B/tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc
-8.2.0/spack-build/prev-x86_64-pc-linux-gnu/libstdc++-v3/libsupc++
/.libs -I/tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0/sp
ack-build/prev-x86_64-pc-linux-gnu/libstdc++-v3/include/x86_64-pc-
linux-gnu -I/tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0
/spack-build/prev-x86_64-pc-linux-gnu/libstdc++-v3/include -I/sha
red/spack/var/spack/stage/gcc-8.2.0-arepu3ta2xtlboacnuxk3su2h3oqc3
lr/gcc-8.2.0/libstdc++-v3/libsupc++ -L/tmp/wpowell/spack-stage/spa
ck-stage-u_7lic/gcc-8.2.0/spack-build/prev-x86_64-pc-linux-gnu/lib
stdc++-v3/src/.libs -L/tmp/wpowell/spack-stage/spack-stage-u_7lic/
gcc-8.2.0/spack-build/prev-x86_64-pc-linux-gnu/libstdc++-v3/libsup
c++/.libs -no-pie -g -O2 -gtoggle -DIN_GCC -fno-exceptions -
fno-rtti -fasynchronous-unwind-tables -W -Wall -Wno-narrowing -Wwr
ite-strings -Wcast-qual -Wmissing-format-attribute -Woverloaded-vi
rtual -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlengt
h-strings -DHAVE_CONFIG_H -static-libstdc++ -static-libgcc -o c
c1plus
9723 cp/cp-lang.o c-family/stub-objc.o cp/call.o cp/class.o cp/c
onstexpr.o cp/constraint.o cp/cp-gimplify.o cp/cp-objcp-common.o c
p/cp-ubsan.o cp/cvt.o cp/cxx-pretty-print.o cp/decl.o cp/decl2.o c
p/dump.o cp/error.o cp/except.o cp/expr.o cp/friend.o cp/init.o cp
/lambda.o cp/lex.o cp/logic.o cp/mangle.o cp/method.o cp/name-look
up.o cp/optimize.o cp/parser.o cp/pt.o cp/ptree.o cp/repo.o cp/rtt
i.o cp/search.o cp/semantics.o cp/tree.o cp/typeck.o cp/typeck2.o
cp/vtable-class-hierarchy.o attribs.o incpath.o c-family/c-common.
o c-family/c-cppbuiltin.o c-family/c-dump.o c-family/c-format.o c-
family/c-gimplify.o c-family/c-indentation.o c-family/c-lex.o c-fa
mily/c-omp.o c-family/c-opts.o c-family/c-pch.o c-family/c-ppoutpu
t.o c-family/c-pragma.o c-family/c-pretty-print.o c-family/c-seman
tics.o c-family/c-ada-spec.o c-family/c-ubsan.o c-family/known-hea
ders.o c-family/c-attribs.o c-family/c-warn.o c-family/c-spellchec
k.o i386-c.o glibc-c.o cc1plus-checksum.o libbackend.a main.o libc
ommon-target.a libcommon.a ../libcpp/libcpp.a ../libdecnumber/libd
ecnumber.a libcommon.a ../libcpp/libcpp.a ../libbacktrace/.libs/
libbacktrace.a ../libiberty/libiberty.a ../libdecnumber/libdecnumb
er.a -L/shared/spack/opt/spack/linux-rhel6-x86_64/gcc-4.8.2/isl-0
.18-xmai3obl46gqlvhruvm4hk7pxa32rbze/lib -lisl -L/shared/spack/opt
/spack/linux-rhel6-x86_64/gcc-4.8.2/gmp-6.1.2-srovmzcugjw3lzgjodl2
ecaw6zbsspx7/lib -L/shared/spack/opt/spack/linux-rhel6-x86_64/gcc-
4.8.2/mpfr-3.1.6-hlkt4pyx7novqzfr3frcb6im5stqjecu/lib -L/shared/sp
ack/opt/spack/linux-rhel6-x86_64/gcc-4.8.2/mpc-1.1.0-ifakhuo2jsdfw
f3hu7hvqj4lbt5tmw2p/lib -lmpc -lmpfr -lgmp -rdynamic -ldl -L/shar
ed/spack/opt/spack/linux-rhel6-x86_64/gcc-4.8.2/zlib-1.2.11-7z2qze
n7xxbgrxy5mijexrtcbtir6bwq/lib -lz
9724 echo | /tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0/spack
-build/./gcc/xgcc -B/tmp/wpowell/spack-stage/spack-stage-u_7lic/gc
c-8.2.0/spack-build/./gcc/ -E -dM - |
9725 sed -n -e 's/^#define ([^][a-zA-Z0-9])./\1/p'
9726 -e 's/^#define (_[^A-Z][a-zA-Z0-9])./\1/p' |
9727 sort -u > tmp-macro_list
9728 /tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0/spack-build/
./gcc/cc1: error while loading shared libraries: libisl.so.15: can
not open shared object file: No such file or directory
9729 /bin/sh /shared/spack/var/spack/stage/gcc-8.2.0-arepu3ta2xtlboacnu
xk3su2h3oqc3lr/gcc-8.2.0/gcc/../move-if-change tmp-macro_list macr
o_list
9730 echo timestamp > s-macro_list
9731 rm -rf include-fixed; mkdir include-fixed
9732 chmod a+rx include-fixed
9733 if [ -d ../prev-gcc ]; then
9734 cd ../prev-gcc && \
...
9825 done
9826 rm -f include-fixed/README
9827 cp /shared/spack/var/spack/stage/gcc-8.2.0-arepu3ta2xtlboacnuxk3su
2h3oqc3lr/gcc-8.2.0/gcc/../fixincludes/README-fixinc include-fixed
/README
9828 chmod a+r include-fixed/README
9829 echo timestamp > stmp-int-hdrs
9830 /tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0/spack-build/
./gcc/xgcc -B/tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0
/spack-build/./gcc/ -xc -nostdinc /dev/null -S -o /dev/null -fself
-test=/shared/spack/var/spack/stage/gcc-8.2.0-arepu3ta2xtlboacnuxk
3su2h3oqc3lr/gcc-8.2.0/gcc/testsuite/selftests
9831 /tmp/wpowell/spack-stage/spack-stage-u_7lic/gcc-8.2.0/spack-build/
./gcc/cc1: error while loading shared libraries: libisl.so.15: can
not open shared object file: No such file or directory
9832 make[3]: *** [s-selftest-c] Error 1
9833 rm gfortran.pod gcc.pod
9834 make[3]: Leaving directory /tmp/wpowell/spack-stage/spack-stage-u
_7lic/gcc-8.2.0/spack-build/gcc' 9835 make[2]: *** [all-stage2-gcc] Error 2
9836 make[2]: Leaving directory /tmp/wpowell/spack-stage/spack-stage-u
_7lic/gcc-8.2.0/spack-build'
9837 make[1]: *** [stage2-bubble] Error 2
9838 make[1]: Leaving directory `/tmp/wpowell/spack-stage/spack-stage-u
_7lic/gcc-8.2.0/spack-build'
9839 make: *** [all] Error 2
See build log for details:
/shared/spack/var/spack/stage/gcc-8.2.0-arepu3ta2xtlboacnuxk3su2h3oqc3lr/gcc-8.2.0/spack-build.out```
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#11396?email_source=notifications&email_token=AAOVY57SOFSUJMWBU7ZW53TPWQF3DA5CNFSM4HLTVACKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV4G4WY#issuecomment-494431835>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAOVY52YLISKRX37D6QTZR3PWQF3DANCNFSM4HLTVACA>
.
|
|
Still searching for a fix, but wanted to tell you what I tried. Did yum install libisl-dev Then was greeted with No package libisl-dev available. so now will look for a repo that has it since mine obviously doesnt. |
|
What OS are you using?
…On Tue, May 21, 2019 at 12:37 PM liamrpowell ***@***.***> wrote:
Still searching for a fix, but wanted to tell you what I tried.
Did yum install libisl-dev
Then was greeted with
No package libisl-dev available.
so now will look for a repo that has it since mine obviously doesnt.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#11396?email_source=notifications&email_token=AAOVY55KV32BAPZKAETFYHTPWQQL7A5CNFSM4HLTVACKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV4PLQQ#issuecomment-494466498>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAOVY52B5FBULTCFDJQ7NXTPWQQL7ANCNFSM4HLTVACA>
.
|
|
RHEL 6.4 on the login node I am using. |
|
RHEL 6 is ancient; to the extent that a number of basic things that are OK
on other systems need to be upgraded / built to make any modern software
work. A simple fix might be to try to install an older version of GCC.
Not sure which GCC you need, I'm still using 4.9.3.
Did you try `yum install libisl-devel`? It looks to me like that's the
standard "development" suffix for RHEL6 packages. Maybe this will help?
http://rpm.pbone.net/index.php3/stat/4/idpl/54411855/dir/centos_other/com/libisl-devel-0.18-9.94.el7.x86_64.rpm.html
…-- Elizabeth
On Tue, May 21, 2019 at 12:44 PM liamrpowell ***@***.***> wrote:
RHEL 6.4 on the login node I am using.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#11396?email_source=notifications&email_token=AAOVY5YLKXL5LK3OHRAGLRTPWQREXA5CNFSM4HLTVACKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV4P5NY#issuecomment-494468791>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAOVY53CPFK6UAVLDJHU533PWQREXANCNFSM4HLTVACA>
.
|
|
I tried Yum install libisl-devel. |
|
Ok did |
I get the impression from https://stackoverflow.com/questions/11912878/gcc-error-gcc-error-trying-to-exec-cc1-execvp-no-such-file-or-directory that the system compiler you are using may need some additional utilities installed to compile a new version of GCC- it is suggesting
|
|
@liamrpowell I haven't seen a response to #11396 (comment) for a few months so I am going to close this. If you can still replicate the issue feel free to reopen the bug (in which case could you try the suggestion at #11396 (comment)?) |
I have Python 3.7.3
When I try spack install py-matplotlib
I get
"py-matplotlib requires python version 3.5:, but spec asked for 2.7.15"
If I try to install using
spack installl py-matplotlib ^python@3.7.3
I get this error.
640 /opt/rh/devtoolset-2/root/usr/libexec/gcc/x86_64-redhat-linux/4.8.2
/ld: /usr/local/lib/libpython3.7m.a(bytesobject.o): relocation R_X8
6_64_32 against `_Py_NoneStruct' can not be used when making a shar
ed object; recompile with -fPIC
641 /usr/local/lib/libpython3.7m.a: could not read symbols: Bad value
I am new to spack and any help would be appreciated.
The text was updated successfully, but these errors were encountered: