Miro

Stateful Installation Failure: Some packages try to install into /opt/ohpc/pub on compute nodes

1

Some packages that are installed on the nodes install files into /opt/ohpc/pub/, which is read-only. This causes installation failure in xCAT stateful installs, where installation happens on running system. For...

ndoutils needs nagios from EPEL?

1

Noticed that ndoutils the dependency is defined as Requires: nagios >= 4 which brings in nagios from EPEL rather that using nagios-ohpc. Is this intentional or an oversight?

python36 getting pulled in for python-Cython

1

Python-Cython requires /usr/bin/python3, which is now provided by python36 rather than python34. So, users end up with mix of python34 and 36. This is related to issue #953

Packages that do not have ohpc string in them

4

Hello, I have noticed that there are rpm packages that do not contain ohpc string in their name. These are: pesign-obs-integration pesign-obs-integration-debuginfo lustre-iokit

Obsolete ( 7 years old) gcc compiler included in pdtoolkit

6

Maybe I am misreading this, but when running OpnHPC packages through a security scanner, an old version of gcc is found in pdtoolkit. This seems to indeed be the case:...

File conflict when installing slurm on SUSE 12.3

2

Following the instructions (using local repository, BTW), I hit file conflict when installing slurm on SUSE 12.3: zypper -n install ohpc-slurm-server Loading repository data... Reading installed packages... Resolving package dependencies......

variable_update=parameter_server fails with XLA in distributed mode

4

Turning on XLA (--xla_compile=True) in distributed mode causes failure during initialization: > 2019-01-08 01:10:13.755644: E tensorflow/core/distributed_runtime/master.cc:315] CreateSession failed because worker /job:worker/replica:0/task:1 returned error: Unavailable: OS Error > Additional GRPC error...

LlaMa2-70b run_accuracy.sh issue with consolidate_results.py

2

LlaMa2-70b has a run_accuracy.py script that as a last step that calls another script, consolidate_results.py. That last step does not work, it expects to read output pkl files, but those...

MigraphX version of reference code

1

Adding MIGraphX version of the BERT reference code for use on AMD GPUs that support ROCm. Credit to Zixian Wang from UCSD for helping validating and testing the code as...

Fix link in GettingStarted.ipynb

1

Replacement for PR #1096