On Oct 3, 3:02pm, net...@precedence.co.uk (Stephen Borrill) wrote: -- Subject: Re: too many inodes error from fsck
| On Thu, 26 Sep 2013, Brett Lymn wrote: | > On Wed, Sep 25, 2013 at 07:58:12PM +0000, Christos Zoulas wrote: | >> | >> And this is i386 I guess, because the number of inodes is suspect: | >> | >> ... 0x51e4a0ed ... I guess your only chance is to try to fix it | >> with 64 bit fsck because given the number of inodes the 32 bit fsck | >> will need to allocate more than 4G... Or edit pass1.c and clamp | >> the inospace and hope of the bast. | >> | > | > Or be really brutal and use fsdb and clri that inode - accepting the | > fact that you will lose data. | | I did a bit of all of this. | | I switched ot 6.1.2 NetBSD/amd64 and used clri to remove the DUP and BAD | inodes. After that, fsck died with out of swap (there was none, as I was | using the USB boot image). So I added swap and fsck eventually died with: | | bad inode number 389260032 to nextinode I think you need to clri that one too... | This is a 7.2TB RAIDframe array. Have used this size of array to >100% | full on netbsd-5 without problems in the past. Something corrupted it... Badly... christos
