Checkpoint and recovery does not work

Create issue
Issue #60 closed
Ian Hinder created an issue

This issue was discussed on the mailing list (8 October 2010 21:55:08 GMT+02:00) but never made it into a bug report. I have reproduced the problem on Damiana and Queen Bee. If I submit a simulation and let it run and make some checkpoint files, and then submit the same simulation again, a second restart is correctly created, but I get the warning on startup

WARNING[L3,P0] (IOUtil): No HDF5 checkpoint files with basefilename 'checkpoint.chkpt' found in recovery directory .

and the simulation starts from iteration 0 rather than recovering.

It looks like simfactory is not hardlinking the checkpoint files from the old restart to the new restart. Being able to use checkpoint/recovery is absolutely necessary for production work.

Keyword:

Comments (2)

  1. Log in to comment