Hi Several people have stated that they have played with a patchless server in new kernels. There is unfortunately an issue with that (even if it proves to be stable, which we are not certain about). To force failover the server currently depends on a special block device setting which basically discards I/O while the server is tearing things down. We will in due course probably do this differently, but at the moment failing over patchless servers is not completely safe. This being said we are very keen to hear how well patchless servers are hodling up, particularly in production environments. - Peter - -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.clusterfs.com/pipermail/lustre-discuss/attachments/20060527/e82e07bf/attachment-0001.html