CIFTS-related Software Download Page

This page contains information and download links to specifications, documents and software affiliated with the CIFTS project.

The FTB API specification

The CIFTS team has proposed an interface specification to allow various components to exchange fault-related information using the Fault Tolerance Backplane. This interface is called as the Fault Tolerance Backplane API (FTB-API).

The FTB API - version 0.5 is the first publicly released version of the FTB-API specification.

Document	Download
FTB-API 0.5 specification	[pdf]

THE FTB SOFTWARE

The CIFTS team has made available an implementation of the FTB-API Version 0.5 specification. Named FTB software, the latest releases of this implementation can be downloaded using the links below.

Software	Download	Platforms Supported
ftb-0.6.2 (latest release)	[tgz]	Linux, Cray, IBM Blue Gene/P
ftb-0.6.1	[tgz]	Linux, Cray, IBM Blue Gene/L

FTB MPI Standard Fault Events Specification

The CIFTS team is standardizing a generic list of fault events that would be common and standardized across all Message Passing Interface (MPI) implementations. Version 1.0 of this specification was presented at SC'10.

Document	Download
FTB MPI 1.0 specification	[pdf]

LIST OF FTB-ENABLED SOFTWARE

Several popular software that are routinely used on high-end computing systems support the FTB-API specification. The below table gives details of (a) FTB-enabled software name and its latest version, (b) Supported FTB-API specification version, (c) Download link for the software (d) List of fault events that the software can publish to the Fault Tolerance Backplane, (e) Contact information

Software	FTB-API specification version	Download Software	Supported Fault Events	Contact
MPICH2 - version 1.3.2p1 (website)	FTB-API VERSION 0.5	[http]	[http]	[Email]
InfiniBand Monitoring Software (FTB-IB) - version 1.0 (website)	FTB-API VERSION 0.5	[http]	[http]	[Email]
MVAPICH-2.1.6 (website)	FTB-API VERSION 0.5	[http]	[http]	[Email]
OPEN MPI-1.5.2 (website)	FTB-API VERSION 0.5	[http]	[http]	[Email]
Berkeley Checkpoint/Restart for Linux (BLCR) - version 0.8.2 (website)	FTB-API VERSION 0.5	[http]	[pdf] (Check README.FTB file in BLCR package)	[Email]