Yaroslav Halchenko

This is "inspired" by the problem originally reported in https://github.com/ReproNim/reproman/pull/438#issuecomment-522047558 with a proposed fix (closed without merge) in https://github.com/ReproNim/reproman/pull/451 to just let `tar` ignore disappearing files. Although we still do...

cannot login to ec2 instance (NITRC CE 5.0) in west-2 zone

2

``` (git)lena:~/proj/repronim/reproman-master[master] $> reproman create -t aws-ec2 my-nitrc-west-2-2 -b region_name=us-west-2 image=ami-02faf30d364b36946 instance_type=t2.medium You did not specify an EC2 SSH key-pair name to use when creating your EC2 environment. Please enter...

create -t aws-ec2 which fails should remove the key it generated

1

Otherwise it requires to breed more of them for no good reason. ```shell (git)lena:~/proj/repronim/reproman-master[master]git $> reproman create -t aws-ec2 my-nitrc-west-2 -b region_name=us-west-2 image=ami-080483fe6f1382eab instance_type=t2.medium You did not specify an EC2...

create -t aws-ec2 should probably make system wide ssh aware of that host/fingerprint

1

ATM after we create aws-ec2 instance, if we decide to use it via datalad (e.g. within `reproman run`) it would use regular ssh (not reproman's login/paramiko) and then upon initial...

run: option to specify resource actions upon at the initiation and upon completion of the run in --follow mode

1

AWS EC2 time is not cheap, we should make it possible to "fire and forget" a computation without monitoring the progress to just shut it down to not incur charges...

pyout: "Interference" between pyout output and interactive prompt

3

Observed while testing #399: If upon `--refresh` access to the resource requires interactive session (e.g. enter a password), then 1. It might be not 100% clear to which resource that...

reproman run with datalad* orchestrators should check versions/extensions

2

We have tried to run some computation on an elderly NITRC AMI (ami-0a4cf63709e153111) with datalad-pair-run orchestrator and container job parameter. It tried, but didn't work and according to stderr it...

- log some details at the DEBUG level (e.g. that it is about to start an instance on ami blah in zone bleh) - issue a warning if no credentials...

I guess we would need only an identity id for that instance, and somehow establish ssh keypair to access it