Packages for Lustre 1.0.4 are now available at the usual location:
http://www.clusterfs.com/download.html
This maintenance release incorporates fixes for a number of issues that
may be visible to Lustre users:
- a fix for some memory allocation problems has been made;
- an interaction between write and truncate, which could result in
truncated data, has been repaired;
- mkfsoptions will work correctly for OSTs (thanks to Jacques-Charles
Lafoucriere for the reports and initial patches);
- a rare race, which could result in a dead client not being properly
evicted for a failure to respond to a lock revocation, has been
fixed;
These caveats still apply to the 1.0.x series:
- running a client and OST on the same node is still not 100% stable.
Application or system hangs are possible.
- If you run out of free space, the "out of space" errors may be
silently eaten by the write cache, and not communicated to the
application. The only workaround is to avoid running out of free
space.
As usual, we strongly encourage you to read the cautions on the download
page carefully. We know that this release of Lustre can be used with
considerable success and excellent performance, but we expect problems
to arise when deployed by people unfamiliar with Lustre. We welcome
reports of problems at http://bugzilla.lustre.org, and on the
lustre-discuss mailing list:
lustre-discuss@lists.clusterfs.com
For those who desire additional support or guaranteed response times, a
variety of support arrangements are also available from CFS.
For the Lustre team,
Mike