Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!mips!wyse!bob From: bob@wyse.wyse.com (Bob McGowen Wyse Technology Training) Newsgroups: comp.unix.xenix Subject: Re: Disk full Keywords: inodes fsck -s Message-ID: <2437@wyse.wyse.com> Date: 19 Sep 89 22:53:01 GMT References: <9359@attctc.Dallas.TX.US> Sender: news@wyse.wyse.com Reply-To: bob@wyse.UUCP (Bob McGowen Wyse Technology Training) Distribution: usa Organization: Wyse Technology Lines: 39 In article <9359@attctc.Dallas.TX.US> tensor@attctc.Dallas.TX.US (Steven Murphy) writes: >Howdy all, > > Every now and then, one of our customers' systems will start >displaying "no space on /dev/xxx (xxx designating the hard disk-I 've >forgotten the exact designation :-(. ). It apparently is *not* due to >the actual disk being full, but that the free list of inodes is full. >Running fsck -s will get the free space back, but on one occasion with >disastrous results (database trashed). What is the explanation for running >out of inodes? Is there something that can prevent it's occurrence. Is there >a place in TFM that explains it, if so help me out. The problem has occurred >on different hardware and Xenix286 revisions, but the last time it happened, >it was with a compaq 386 and Xenix286 2.2.3. > Thanks, > Steve You have one file per inode, and the number of files = number of inodes. Check under mkfs for more info. In particular, the following form allows you to increase the number of inodes available. mkfs /dev/name number_of_blocks:number_of_inodes This form works with 2.2.x and 2.3.x XENIXs (according to the man page, I have not used this form under either version). I do not know about earlier versions though I expect not. The blocks are BSIZE, which is 1K, so you can figure the number to use from the df output by dividing by 2 (df reports in 512 byte block size). I believe that the "standard" number of inodes is approximately 25% of the number of blocks, so adjust your number_of_inodes up accordingly. Remember that you will "loose" blocks for storage of the inodes, but there are 64 inodes per block so the "loss" is not high. I hope this is helpful. Bob McGowan (standard disclaimer, these are my own ...) Customer Education, Wyse Technology, San Jose, CA ..!uunet!wyse!bob bob@wyse.com