About one month ago we created a new 30GB filesystem under OCFS2 1.4.4 and shared it between two nodes. The filesystem was created with "-T mail" and is used as a staging area for small JPEG photos and this means that upwards of 15,000 files are created and/or deleted on a daily basis. This weekend we hit the erroneous "our of space" error, even though the filesystem was only 53% full and so I decided to upgrade to the latest release of OCFS2. Even after the upgrade, we still couldn't create new files on this filesystem and so I backed it up, recreated the filesystem and restored the data. As a result of how we use this filesystem, it occurred to me that fragmentation was very likely to recur in the near future :-( Reading the release notes for OCFS2 1.4.7, it states that the fix allows a node to steal free blocks from another node's extent allocator. What I have done is to allocate 16 node slots, even though there will never be more than three nodes accessing this filesystem. My intention is to give OCFS2 plenty of nodes from which to steal these free bocks - it this a reasonable precaution given our usage? Will doing this make any difference? Your advice would be greatly appreciated. Regards, Mick. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20100510/fa2f683c/attachment.html