From: Theodore Tso Subject: Re: EXT4 ENOSPC Bug Date: Tue, 2 Dec 2008 15:33:15 -0500 Message-ID: <20081202203315.GA20858@mit.edu> References: <200811291418.24672.andres@anarazel.de> <200812021559.05103.andres@anarazel.de> <20081202164709.GC18162@mit.edu> <200812021847.35771.andres@anarazel.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Andreas Dilger , LKML , linux-ext4@vger.kernel.org To: Andres Freund Return-path: Received: from www.church-of-our-saviour.org ([69.25.196.31]:37728 "EHLO thunker.thunk.org" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1751018AbYLBUdV (ORCPT ); Tue, 2 Dec 2008 15:33:21 -0500 Content-Disposition: inline In-Reply-To: <200812021847.35771.andres@anarazel.de> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Tue, Dec 02, 2008 at 06:47:24PM +0100, Andres Freund wrote: > > I think that I have seen the problem on metadata only changes (find > /tmp -type f|xargs touch) as well, but sometimes metadata changes > were possible while file creation was not. Hmm... really? Was the error message ENOSPC, or something else? The only way I can see that it might be ENOSPC would be if we got an error in the jbd2 layer, in start_this_handle(), but that would have resulted in a kernel printk: if (nblocks > journal->j_max_transaction_buffers) { printk(KERN_ERR "JBD: %s wants too many credits (%d > %d)\n", current->comm, nblocks, journal->j_max_transaction_buffers); ret = -ENOSPC; goto out; } If that wasn't too long ago, you might want to search your old system log files in /var/log for any "JBD" or "ext4" messages. That might be an important clue, although that seems rather unlikely to me. Or in the case where metadata changes were failing, perhaps they were doing so with some other error that ENOSPC? - Ted