ESGF replication¶
Introduction¶
This document describes how to untar an ESGF archives partial mirror on a local cluster and keep it up to date.
Synda installation¶
See synda installation.
Create a selection file to describe which data to replicate¶
See the selection file section
Note
You can untar many selection files (e.g. one per project). Selection file(s) must be stored in the “selection” folder.
File discovery¶
Install your selection file:
synda install -s <selection-file>
Or upgrade the file discovery:
synda upgrade
At this point, files metadata are stored in local database and data download can begin.
Files download¶
Watching download progress¶
Commands below give download status and progress:
synda queue
synda watch
Update datasets to the latest version¶
In ESGF, a dataset can change over time and thus have several versions. Re-run the discovery to search for new versions:
synda upgrade
Then start the Synda service to start the download of new versions if any.
Error management¶
Errors can occur during download for different reasons: remote server is down, incorrect files access right, authorization issue, authentication issue, etc.
Retrying download in errors¶
Command below can be used to retry transfer(s) in error:
synda retry
Note
crontab can be used to retry errors (e.g. every day).
Changing replica for all files in errors¶
Sometime, current file replica always fails as the remote server is down. One way to solve the problem is to change the file replica.
To change the replica for all files in error, use command below:
synda replica next
Getting more details about errors¶
Log files below contain useful information about errors:
discovery.log
contains information regarding discovery.transfer.log
contains information regarding download.debug.log
contains low-level download information.
Note
Log files are stored in $HOME/log
folder.