OK, got some more info. If the file system is a multiple of the AU size, then VxFS can hit this condition. So, in terms of future avoidance >>always create file systems with 1 block more than a multiple of 2^15 blocks >>(and apply the same restriction when growing or shrinking file systems), then >>we'll always have enough hidden space to grow the fs. Note: that's file system block size which is 1k (default), 2k, 4k, or 8k. My source also offered some possibilities on why even removing files didn't allow them to grow. 1. this may be because we need to allocate 8k extents as part of growing an fs and the files he removed didn't have any 8k or larger extents.
(if that's the case then he would be able to make space to grow by removing files with larger extents*.) I'm sorry I didn't think of his second suggestion immediately, as it happens not infrequently: 2. or maybe he had a checkpoint and the blocks for the files he removed just got transferred to the ckpt, in which case he would need to remove the ckpt *You can get some extent information for an individual file from fsadm_vxfs, e.g.,: $ <mailto:[EMAIL PROTECTED]</export/data/tpm/customers>$> fsadm -t vxfs -E -f mynovel.txt Extent Fragmentation Report Total Average Average Total Files File Blks # Extents Free Blks 1 84 4 211974585 ... The free blocks information is for the file system as a whole. With 84 blocks and 4 extents, one of them must be at least 8k in this example. (the above example is -t for linux; use -F for Solaris|HPUX or -V for AIX) Regards, Scott ________________________________ From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Scott Kaiser Sent: Monday, March 17, 2008 1:29 PM To: Rajiv Gunja; Sengor Cc: Asim Zuberi; veritas-vx@mailman.eng.auburn.edu; Doug Hughes Subject: Re: [Veritas-vx] vxfs filesystem full - can not expand. Inodes are dynamically allocated in VxFS, so that's not likely to be the cause (unless you have a multiple billion inodes or a very, very old disk layout). Enhancements were a couple versions ago to reserve space internally within the file system so users wouldn't have to manually reserve free space. See: http://seer.entsupport.symantec.com/docs/249933.htm But you said you're running on version 5. So it sounds like you've hit one of the 'rare cases' What is the exact size of the file system? I seem to recall the rare cases involved being a multiple of some number. Regards, Scott ________________________________ From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Rajiv Gunja Sent: Monday, March 17, 2008 11:10 AM To: Sengor Cc: Asim Zuberi; Doug Hughes; veritas-vx@mailman.eng.auburn.edu Subject: Re: [Veritas-vx] vxfs filesystem full - can not expand. -- That is very odd. How are you doing on i-nodes? Too many / any free left? -- RM fails, how about mv (move) there should be a way to free up i-nodes and or space. -GGR On Sun, Mar 16, 2008 at 11:31 PM, Sengor <[EMAIL PROTECTED]> wrote: On 3/17/08, Asim Zuberi <[EMAIL PROTECTED]> wrote: > Did you consider doing "fsck -F vxfs" on the filesystem? Nope, since we need the fs to remain online. -- _________________________________/ sengork.blogspot.com /¯¯¯¯
_______________________________________________ Veritas-vx maillist - Veritas-vx@mailman.eng.auburn.edu http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx
