tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: FS corruption because of bufio_cache pool depletion?

> raid1: IO failed after 5 retries.
> [...]
> Does that ring a bell with anyone?

> Greg Oster suggested to prime the bufcache_pool and see what the effect is.
He suggested something similar to me when I had that strange problem of 
RAIDframe failing without a device failure.

He also analysed my problem as follows:

> To make a long story short, rf_CreateDiskQueueData() is being called
> with PR_NOWAIT, and so if it is unable to get items from the pools
> what happens then is that something like this gets called:
>               (node->wakeFunc) (node, ENOMEM);
> which ends up eventually marking the request as 'bad' and counts it 
> as one IO try.  If after 5 tries in total it can't get the memory it 
> needs, it returns EIO.  Where you hit the panics is that some things 
> (softdeps) don't deal well with EIO on critical blocks :-}
> So now to figure out how to fix the problem....

Any news about that?

Home | Main Index | Thread Index | Old Index