update some man pages

man/scr.1.in

@@ -1,18 +1,21 @@
 .TH scr 1 "@META_DATE@" "@META_ALIAS@" "@META_NAME@"
 
 .SH DESCRIPTION
-The Scalable Checkpoint / Restart (SCR) library provides an interface that
-codes may use to write out and read in application-level checkpoints in a
-scalable fashion.  In the current implementation, checkpoint files are
-cached in local storage (hard disk or RAM disk) on the compute nodes.
-This technique provides scalable aggregate bandwidth and uses storage
-resources that are fully dedicated to the job.  This approach addresses
-the two common drawbacks of checkpointing a large-scale application to
-a shared parallel file system, namely, limited bandwidth and file system
-contention.  In fact, on current platforms, SCR scales linearly with the
-number of compute nodes.  It has been benchmarked as high as 720GB/s
-on 1094 nodes of Atlas, which is nearly two orders of magnitude faster
-than the parallel file system.
+The Scalable Checkpoint / Restart (SCR) library enables MPI applications
+to utilize distributed storage on Linux clusters to attain high file I/O
+bandwidth for checkpointing and restarting large-scale jobs. With SCR,
+jobs run more efficiently, recompute less work upon a failure, and reduce
+load on critical shared resources such as the parallel file system.
+
+SCR caches checkpoint files in storage local to the compute nodes, and it
+applies a redundancy scheme such that files can be recovered in the event
+of a failure. When a failure occurs, the current run is killed. If there
+are sufficient spare nodes and time remaining in the resource allocation,
+another run may be restarted from the cached checkpoint files. Otherwise,
+the cached checkpoint files are flushed to the parallel file system, and
+the resource allocation is released. With this approach, file I/O
+bandwidth scales linearly with the number of compute nodes. On large
+clusters, SCR is often 100x to 1000x faster than the parallel file system.
 
 .SH SEE ALSO
 \fISCR User Manual\fR

man/scr_index.1.in

@@ -1,12 +1,12 @@
 .TH scr_index 1  "@META_DATE@" "@META_ALIAS@" "@META_NAME@"
 .SH NAME
-scr_index \- manages status of checkpoints stored on parallel file system
+scr_index \- manages status of datasets stored on parallel file system
 
 .SH SYNOPSIS
 .B "scr_index [options]"
 
 .SH DESCRIPTION
-The \fBscr_index\fR command manages checkpoints stored on the parallel
+The \fBscr_index\fR command manages datasets stored on the parallel
 file system.
 
 .SH OPTIONS
@@ -15,10 +15,13 @@ file system.
 List contents of index (default behavior).
 .TP
 .BI "-a, --add " DIR
-Add checkpoint directory to index.
+Add dataset directory to index.
 .TP
 .BI "-r, --remove " DIR
-Remove checkpoint directory from index.  Does not delete corresponding files.
+Remove dataset directory from index.  Does not delete corresponding files.
+.TP
+.BI "-c, --current " DIR
+Specify directory as restart directory.
 .TP
 .BI "-p, --prefix"
 Specify prefix directory (defaults to current working directory).
@@ -28,39 +31,38 @@ Print usage.
 
 .LP
 When adding a directory to the index, the command checks
-that the checkpoint directory on the parallel file system contains a
-complete and valid set of checkpoint files, it rebuilds missing
-checkpoint files if possible, and it writes an \fIscr_summary.txt\fR
+that the dataset directory on the parallel file system contains a
+complete and valid set of files, it rebuilds missing
+files if possible, and it writes a \fIsummary\fR
 file, which is used by the SCR library to fetch and redistribute files
 to appropriate ranks upon a restart.
 
 One may invoke the command outside of a SLURM job allocation, which is
-useful to check and rebuild a checkpoint set in which \fBscr_postrun\fR
+useful to check and rebuild a dataset set in which \fBscr_postrun\fR
 may have failed to complete its internal call to \fBscr_index\fR.
 
-When listing checkpoints, the internal SCR checkpoint id is shown,
-along with the checkpoint directory name.  In addition, the time
-the checkpoint was flushed is shown as well as a set of flags.
-The flags are shown as a series of columns consisting of single characters.
-The first column represents whether the checkpoint is complete.
-An 'x' in the first column implies the checkpoint is incomplete while 'c' means the checkpoint is complete.
-An 'f' in the second column indicates that SCR has marked the checkpoint as failed.
-During a restart, SCR will not attempt to fetch a checkpoint that is marked as incomplete or failed.
+When listing dataset, the internal SCR dataset id is shown,
+along with a flag denoting whether the dataset is valid,
+the time the dataset was flushed, and the dataset directory name.
+The dataset marked as \fBcurrent\fR is denoted with a * symbol.
+During a restart, SCR will attempt to fetch the most recent
+checkpoint starting from the current dataset.
+SCR will not attempt to fetch any dataset marked as invalid.
 
 .SH EXAMPLES
 .TP
-(1) List checkpoint direcotries in index file:
+(1) List dataset direcotries in index file:
 .nf
 >> scr_index --list
-FLAGS  FLUSHED              CKPT  DIRECTORY
-c-     2011-04-11T17:59:27    18  scr.2011-04-11_17:59:27.427417.18
-c-     2011-04-11T17:59:24    12  scr.2011-04-11_17:59:24.427417.12
-c-     2011-04-11T17:59:21     6  scr.2011-04-11_17:59:21.427417.6
+   DSET VALID FLUSHED             DIRECTORY
+*    18 YES   2015-09-28T16:46:22 scr.dataset.18
+     12 YES   2015-09-28T16:43:40 scr.dataset.12
+      6 YES   2015-09-28T16:43:02 scr.dataset.6
 .fi
 .TP
-(2) Add checkpoint directory to index file, rebuild missing checkpoint files if necessary, and create summary file:
+(2) Add dataset directory to index file, rebuild missing files if necessary, and create summary file:
 .nf
->> scr_index --add scr.2008-10-20_14:22:10.23167.50
+>> scr_index --add scr.dataset.20
 .fi
 
 .SH SEE ALSO