Skip to content

Commit

Permalink
Merge pull request #46 from DouglasNeuroInformatics/fix_37
Browse files Browse the repository at this point in the history
  • Loading branch information
gdevenyi committed Aug 20, 2021
2 parents 6ecc03d + f1cd846 commit c2a4f80
Show file tree
Hide file tree
Showing 2 changed files with 54 additions and 18 deletions.
16 changes: 3 additions & 13 deletions docs/about_the_platform/index.md
Original file line number Diff line number Diff line change
Expand Up @@ -24,17 +24,7 @@ and OpenCL acceleration suitable applications.

The platform workstations run Ubuntu Linux variants, with full productivity
software (LibreOffice, GIMP, Inkscape, etc), modern web browsers (Chrome, Firefox),
and a large suite of scientific software. Scientific software is avaialble in the
and a large suite of scientific software. Servers run Ubuntu Linux server, as well as
compute nodes. Limited Windows workstations are available to provide access to proprietary
software such as E-Prime. Scientific software is available in the
{ref}`software quarantine<using_the_system/access_to_software:accessing scientific software (quarantine aka modules)>`.

## Filesystem layout

Each computer in the system has access to two system-wide network filesystems.
User home directories `$HOME` are mounted from the user server (currently `cicus03`).
Home directories are suitable for storing the regular confiugration files, as well
as papers and similar files, it should not be used to store data. The high performace
filesystem (currently hosted on `cicss03`) is available under the `/data` path.
All users have access to scratch storage at `/data/scratch`, which is suitable
for storing data during processing. Scratch does not keep any historical versions
and is not backed up, so it should not be relied upon for long-term storage. In
the future it is expected a date-based deletion policy will be implemented.
56 changes: 51 additions & 5 deletions docs/using_the_system/access_to_data.md
Original file line number Diff line number Diff line change
@@ -1,17 +1,63 @@
# Access to Data

## Filesystem Layout

Each computer in the system has access to two system-wide network filesystems.
User home directories `$HOME` are mounted from the user server (currently `cicus03`)
on `/home/cic/<username>`. Home directories are suitable for storing the regular configuration files, as well
as papers and similar files, it should not be used to store data.

The high performance filesystem (currently hosted on `cicss03`) is available under the `/data` path.
All users have access to scratch storage at `/data/scratch`, which is suitable
for storing data during processing. Scratch does not keep any historical versions
and is not backed up, so it should not be relied upon for long-term storage. In
the future it is expected a age-based deletion policy will be implemented to
maintain sufficient free space and encourage users to use storage properly.

Paid storage for individual lab groups is also available under the `/data/` path,
the exact name depends upon what the group decided during creation.

```{admonition} Mounting on demand
Network filesystems are not automatically mounted on boot-up on machines, but
rather mounted lazily on-demand. As such, `/data` will appear empty before
attempts are made to access a specific filesystem. GUI file managers do
not work particularly well in this scenario, as there will be no folders
to interact with. It is recommended to perform file management using the
command line tools.
```

## Transferring Data

Access to filesystems is enabled via the `scp/sftp` functionality of `ssh`, as
well as the `rsync` program over `ssh`. For data transfers, please connect
directly to `cicss03` to bypass any round-trip data would need to travel if
performing transfers to workstations. Linux and OSX users can find the `sftp`
well as the `rsync` program over `ssh`. Linux and OSX users can find the `sftp`
`scp` and `rsync` commands in their terminal. Windows users can use
[WinSCP](https://winscp.net/) or [FileZilla](https://filezilla-project.org/)
for a graphical tool to access data.
for a graphical tool to access data or [MobaXterm](https://mobaxterm.mobatek.net/)
for a proper Linux-like terminal.

Here are few example commands for data transfer, note that these commands assume that
you are within the Douglas:

```bash
# Copying a directory of files to cicss03 in /data/scratch/<username>
$ rsync -avz directory <username>@cicss03:/data/scratch/<username>
# Copying another directory from cicss03 to your local system
$ rsync -avz <username>@cicss03:/data/scratch/<username>/another_dir .
```

```{admonition} Resumable file transfers
`rsync` is strongly recommended for all data transfers, as it supports resuming interrupted transfers
```

```{admonition} Bulk data transfer
For bulk data transfers, please connect directly to `cicss03` to bypass any round-trip data would need to travel if
performing transfers to workstations.
```

```{admonition} Real time data access
The [sshfs](https://github.com/libfuse/sshfs) project allows for filesystems to be mounted remotely via ssh. See the site for details and for windows see [here](https://github.com/billziss-gh/sshfs-win).
The [sshfs](https://github.com/libfuse/sshfs) project allows for filesystems to be mounted remotely via ssh.
This allows you to access files without having to explicitly transfer them back-and-forth.
See the site for details and for windows see [here](https://github.com/billziss-gh/sshfs-win).
```

Expand Down

0 comments on commit c2a4f80

Please sign in to comment.