Tests fail, but GYRE works in MESA

[nwtuchow]

Hello,
I've been using GYRE in MESA r10398 within the astero module, but I now want to get GYRE working separately from MESA. I've been following the instructions to get started but when I try to 'make test', I find that one of them fails. I get the following output:

Code: Select all

TEST numerics (rotation, traditional approximation)...
At line 134 of file gyre_r_tar_rot.f90
Internal Error: stash_internal_unit(): Stack Size Exceeded

Error termination. Backtrace:
#0  0x5393f2 in ???
#1  0x4bf464 in ???
#2  0x4e45ad in ???
#3  0x4eeb50 in ???
#4  0x7f319ab138dd in gomp_thread_start
	at /root/mesasdk-src/gcc/libgomp/team.c:120
#5  0x7f319a491e24 in ???
#6  0x7f319a1bf34c in ???
#7  0xffffffffffffffff in ???
HDF5-DIAG: Error detected in HDF5 (1.8.17) thread 0:
  #000: H5G.c line 812 in H5Gclose(): not a group
    major: Invalid arguments to routine
    minor: Inappropriate type
HDF5-DIAG: Error detected in HDF5 (1.8.17) thread 0:
  #000: H5G.c line 812 in H5Gclose(): not a group
    major: Invalid arguments to routine
    minor: Inappropriate type

Any ideas how to resolve these problems and get the tests to pass? GYRE has been working properly for me when it's called in the astero search module, but somehow it is not running passing these tests.

Thanks,
Noah Tuchow

[rhtownsend]

Hello,
I've been using GYRE in MESA r10398 within the astero module, but I now want to get GYRE working separately from MESA. I've been following the instructions to get started but when I try to 'make test', I find that one of them fails. I get the following output:

Code: Select all

TEST numerics (rotation, traditional approximation)...
At line 134 of file gyre_r_tar_rot.f90
Internal Error: stash_internal_unit(): Stack Size Exceeded

Error termination. Backtrace:
#0  0x5393f2 in ???
#1  0x4bf464 in ???
#2  0x4e45ad in ???
#3  0x4eeb50 in ???
#4  0x7f319ab138dd in gomp_thread_start
	at /root/mesasdk-src/gcc/libgomp/team.c:120
#5  0x7f319a491e24 in ???
#6  0x7f319a1bf34c in ???
#7  0xffffffffffffffff in ???
HDF5-DIAG: Error detected in HDF5 (1.8.17) thread 0:
  #000: H5G.c line 812 in H5Gclose(): not a group
    major: Invalid arguments to routine
    minor: Inappropriate type
HDF5-DIAG: Error detected in HDF5 (1.8.17) thread 0:
  #000: H5G.c line 812 in H5Gclose(): not a group
    major: Invalid arguments to routine
    minor: Inappropriate type

Any ideas how to resolve these problems and get the tests to pass? GYRE has been working properly for me when it's called in the astero search module, but somehow it is not running passing these tests.

Thanks,
Noah Tuchow

Hi Noah --

Thanks for the error report. It's worth noting that the self-tests that GYRE does when you install it stand-alone are quite a bit more rigorous than when it is built inside MESA. So, it's possible for it to build and run just fine in MESA, and yet give errors like this when you run 'make test'.

To get started in diagnosing the problem, could you tell me what platform you are running on?

Many thanks,

Rich

[nwtuchow]

I'm running GYRE on a CentOS 7 linux machine

[rhtownsend]

I'm running GYRE on a CentOS 7 linux machine

What processor type, number of cores, and RAM does your machine have?

Also, what version of GYRE are you using? It's possible that this problem can arise on many-core machines with older versions of GYRE.

cheers,

Rich

[nwtuchow]

My computer has an Intel Xeon processor E5-2630 v3 with 8 cores and 16 threads. It has 15 GB of RAM.

I'm running GYRE 5.1

[rhtownsend]

My computer has an Intel Xeon processor E5-2630 v3 with 8 cores and 16 threads. It has 15 GB of RAM.

I'm running GYRE 5.1

Ah -- that's likely your problem. GYRE 5.2, quoting from the release notes, "Fixed various memory errors arising when the traditional approximation is used on a many-core machine (these errors originate from thread-unsafe I/O)".

In light of this, I suggest you upgrade to 5.2.

cheers,

Rich

[nwtuchow]

Alright, GYRE 5.1 seems to be working for the purposes I'm using it for, but if I encounter problems in the future I'll upgrade to 5.2

[rhtownsend]

Alright, GYRE 5.1 seems to be working for the purposes I'm using it for, but if I encounter problems in the future I'll upgrade to 5.2

Fair enough -- it's worth noting that this will only ever be an issue if you're using the traditional approximation of rotation (TAR).

cheers,

Rich