From: Paul H. Hargrove (PHHargrove_at_lbl_dot_gov)
Date: Thu Jan 29 2009 - 11:35:31 PST
Neal Becker wrote:
> I'd like to update my blcr python module. I believe blcr API has changed a
> bit since I wrote that.
>
> Where can I find an example and/or some doc to follow? This is intended for
> very simple checkpointing - single thread, no MPI etc.
>
Neal,
The version you emailed almost exactly 1 year ago uses current BLCR
interfaces. You could update to replace cr_poll_checkpoint() with the
recently added cr_poll_checkpoint_msg() if you want to get a buffer with
the kernel messages on failure and/or warning. You can now pass
CR_CHKPT_ASYNC_ERR in the cr_flags to delay error reporting of most
request-time errors until the poll step. Other than that, I don't think
anything has changed that would affect what you already have.
In 0.8.0 we added library interfaces for requesting restarts, for
which you may add support to your python module.
To answer the original question:
tests/crut.c:crut_checkpoint_to_file() is a good example of
requesting a checkpoint of oneself
util/cr_checkpoint/cr_checkpoint.c:real_main() is the canonical
example for arguments to cr_request_checkpoint() and meanings of errno
values
-Paul
--
Paul H. Hargrove PHHargrove_at_lbl_dot_gov
Future Technologies Group Tel: +1-510-495-2352
HPC Research Department Fax: +1-510-486-6900
Lawrence Berkeley National Laboratory