26 April 2008
Since the GlusterFS v1.2-BENKI Jan 2007 release, we have long awaited
for this day. v1.3 development has been a roller coaster ride. We
tried to achieve too many goals at the same time. After a number of
revisions, we now have a good combination of useful functionalities,
decent performance and acceptable stability. We name this version of
GlusterFS v1.3.8 as the official beginning of stable 1.3-SUSKE series.
Download: http://ftp.gluster.com/pub/gluster/glusterfs/1.3/
Documentation: http://www.gluster.org/docs/index.php/GlusterFS
Supported OS: GNU/Linux, Mac OS X Leopard, FreeBSD (server only),
Solaris (server only).
(Some users got GlusterFS client to work on Solaris and
FreeBSD. We are yet to qualify the availability).
Whats in 1.3-SUSKE
Clustering Mode:
- File level distribution (Unify)
- Automatic file replication / Mirroring (AFR)
- Distributed striping (Stripe)
- Mixed mode (Unify + AFR + Stripe)
- Round robin
- Random
- Wild card
- Adaptive least usage
- Non-uniform file system access
- TCP/IP
- Infiniband verbs RDMA
- Infiniband sockets direct protocol
- Unix domain socket
- Booster (kernel bypass)
- I/O cache
- I/O threads
- Read ahead
- Write behind
- POSIX FS
- *Berkeley DB
- POSIX locks
- Trashcan
- Authentication - user/pass, source-IP
- *Meta (proc like management / monitoring interface)
- Trace logger
- Rot-13 rudimentary encryption
- Filter
- Fixed owner / group
- GlusterFS Unit Test Automation
Known Issues (Top Priority):
- Default timeout of 108 seconds is too long
- BerkeleyDB backend - symlink, rmdir and transactions support pushed to next release
- IB verbs reconnection issue on certain firmware versions
«