HTCondor Deployment Guide: updates

[maany]

• The batch server should have an amount of memory per batch slot for the shadows?

• The copy icon copies everything from its box, not just the command(s) --> put the outputs into their own boxes?

• LC: yum lacks the "-y" option.

• To be removed: mkdir -p /etc/simple_grid/site_config (because it already exists by that time).

• Better anonymize the e-mail addresses etc! E.g. admin@my-host.my-domain.

• Explicitly say that the config master will not appear in site level config?

• Make LC host names less generic? Use names like those of our QA cluster?

• simple-lc-node[01].cern.ch appear twice: we should use the variables instead.

• The runtime variables sections are already present in the template.

• Typo in the template: clustere.

• The lightweight_components template section only has 1 WN vs. 2 in the tutorial.

• An empty line is missing before supported_virtual_organizations.

• Can we supply a supplemental_config example e.g. for augmenting /etc/hosts?

• Why does num_slots not appear in HTCondor-CE config-schema.yaml?

• Example showing how to change the number of accounts for a VO?

• Issues with the pre-compilation phase:

  • A "cd ~" should be done first, to avoid the new directory being created in an awkward place.
  • The pip command is not installed --> yum install python-pip --> but then it says:
  • You are using pip version 8.1.2, however version 20.0.2 is available.
  • pip install --upgrade pip

• I think all commands that can produce a lot of output should have stdout
  and stderr redirected, possibly to different files.

• Has an issue been opened concerning the warnings about "hiera" etc?

• The first bolt command must not have spaces in the node list.

• All bolt commands need to have "--nodes" replaces with "-t" instead.

[maany]

Commands that generate a lot of output will be abstracted in the simple command line tool. Add a warning in the documentation

[maarten-litmaath]

In the template there still is a repository_url to be fixed:

https://github.com/andria009/simple_htcondor_batch

[maany]

Noted. Thanks

[maarten-litmaath]

The template still refers to long variable names like "lightweight_component01_fqdn".

[maany]

Fix site_infrastructure section

Let's say lightweight_component01_fqdn corresponds to simple-lc-node0.cern.ch, while lightweight_component01_ip_address corresponds to 188.184.104.25, then the above section can be re-written as

[maany]

@maarten-litmaath , the main template generated by the puppet module does not include references to the long names.
https://github.com/simple-framework/simple_grid_puppet_module/blob/1.0.x/templates/site_level_config_file.yaml.epp
I have retained the long names in the other template:
https://github.com/simple-framework/simple_grid_puppet_module/blob/1.0.x/templates/cream_torque_sample_site_level_config_file.yaml.epp

Are you referencing to the latter?

[maarten-litmaath]

My bad: I forgot I had already started editing it before doing a diff with the old one, sorry! :-(