Difference between revisions of "Swestore-iRODS-at-NSC"
(→Accessing iRODS from the compute cluster) |
|||
Line 20: | Line 20: | ||
You should have the client configuration files in place under | You should have the client configuration files in place under | ||
− | the <code>.irods</code> directory as it has been described under | + | the <code>~/.irods</code> directory as it has been described under |
the introductory pages. | the introductory pages. | ||
Under your home directory you should create a directory called | Under your home directory you should create a directory called | ||
− | <code>.irods</code> which | + | <code>.irods</code> which will contain the iRODS client configuration |
file. That file is <code>irods_environment.json</code>. It should | file. That file is <code>irods_environment.json</code>. It should | ||
look like: | look like: | ||
Line 68: | Line 68: | ||
To transfer data from the cluster directory it is a good idea to use | To transfer data from the cluster directory it is a good idea to use | ||
<code>irsync</code> like: | <code>irsync</code> like: | ||
− | irsync -r -K local_directory i:irods_directory | + | irsync -v -r -K local_directory i:irods_directory |
where | where | ||
− | -r for recursive copy of the whole directory tree | + | -v for verbose. |
− | -K to compute and check checksums | + | -r for recursive copy of the whole directory tree. |
+ | -K to compute and check checksums. This involves higher CPU usage | ||
+ | but gives confidence for the integrity of the data stored. | ||
i.e.: | i.e.: | ||
irsync -r -K mydir i:/snic.se/projects/projectdir | irsync -r -K mydir i:/snic.se/projects/projectdir |
Revision as of 09:41, 9 December 2020
Swestore iRODS at NSC
iRODS is a distributed storage management system which enables the user to control the access to his/her data via policies which are expressed as rules. Swestore iRODS is hosted on servers at PDC and NSC. We are running version 4.2.6. Basic information about how to store and access data had been described above. Additional information about iRODS can be found at the iRODS web site.
iRODS system at NSC
At NSC we host a replica metadata server and two storage server nodes at the moment. All data stored is replicated between PDC and NSC.
Accessing iRODS from the compute cluster
Data stored in iRODS is accessible from tetralith
.
We have got the command line client installed on the cluster,
iCommands should work from the login node and also from the batch
nodes.
You should have the client configuration files in place under
the ~/.irods
directory as it has been described under
the introductory pages.
Under your home directory you should create a directory called
.irods
which will contain the iRODS client configuration
file. That file is irods_environment.json
. It should
look like:
{ "irods_host": "irods-login.swestore.se", "irods_port": 2432, "irods_default_resource": "pdc-gpfs", "irods_home": "/snic.se/home/<YOUR_SNIC_USER_NAME>", "irods_cwd": "/snic.se/home/<YOUR_SNIC_USER_NAME>", "irods_user_name": "<YOUR_SNIC_USER_NAME>", "irods_zone_name": "snic.se", "irods_client_server_negotiation": "request_server_negotiation", "irods_client_server_policy": "CS_NEG_REFUSE", "irods_encryption_key_size": 32, "irods_encryption_salt_size": 8, "irods_encryption_num_hash_rounds": 16, "irods_encryption_algorithm": "AES-256-CBC", "irods_default_hash_scheme": "SHA256", "irods_match_hash_policy": "compatible", "irods_authentication_scheme": "PAM" }
where the token <YOUR_SNIC_USER_NAME>
should be replaced
with your SNIC user name.
Issue:
module load irods-client/4.2.6
after which you should authenticate using iinit
and then the iCommands like
iget
, iput
, ils
should work as expected.
The ihelp
command will print a short description of the irods commands.
First issue
iinit
which will ask for your iRODS password. You can specify a longer time limit
for the session using the --ttl
option, iinit -h --ttl
will print help using this. The default generally will suffice.
Use
ils
just to check if you can access your home directory. Please use the project
directory to store files, which will be a subdirectory under
//snic.se/projects/<SNIC_PROJECT_NAME>
.
To transfer data from the cluster directory it is a good idea to use
irsync
like:
irsync -v -r -K local_directory i:irods_directory
where
-v for verbose. -r for recursive copy of the whole directory tree. -K to compute and check checksums. This involves higher CPU usage but gives confidence for the integrity of the data stored.
i.e.:
irsync -r -K mydir i:/snic.se/projects/projectdir
If, for whatever reason the transfer had been interrupted the second invocation of the command with the same arguments should pick up and continue from that point.
Detailed description of the command line client can be found at documentation at the iRODS site.