"Nathan Scott" <nathans@xxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> On Nov 30, 8:56am, Thomas Graichen wrote:
>> Subject: Re: alpha again
>> i will wait with this - because due to russels post it seems to
>> happen earlier and not really on umount
> yup, that blew my theory out of the water.
>> just for my understanding: it will only be seen on umount
>> because the trashed blocks will only then be written to disk
>> and we are looking at the on disk layout with xfs_db only
>> - right?
> Russells point was it isn't only seen on unmount though - thats
> just when we happened to look previously - and we weren't doing
> anything inbetween mount & umount which might force the agf out
> earlier (we are now though). yes, xfs_db only looks ondisk.
maybe this helps a bit more: i was now able to reproduce it also
before umount - it takes quite a lot of stuff to copy onto the fs
before i see the first corruption with xfs_db but i can see it now
before umount too ... i think this is due to the agressive caching
technical director innominate AG
clustering & security the linux architects
tel: +49-30-308806-13 fax: -77 http://www.innominate.com