| To: | Andi Kleen <ak@xxxxxxx> |
|---|---|
| Subject: | Re: XFS unhappy with large holey loopback and syncs |
| From: | Steve Lord <lord@xxxxxxx> |
| Date: | Tue, 29 Nov 2005 08:53:17 -0600 |
| Cc: | David Chinner <dgc@xxxxxxx>, linux-xfs@xxxxxxxxxxx |
| In-reply-to: | <20051129091302.GB19515@xxxxxxxxxxxxx> |
| References: | <20051129003611.GF7209@xxxxxxxxxxxxxx> <20051129032937.GZ501696@xxxxxxxxxxxxxxxxx> <20051129091302.GB19515@xxxxxxxxxxxxx> |
| Sender: | linux-xfs-bounce@xxxxxxxxxxx |
| User-agent: | Mozilla Thunderbird 1.0.7-1.1.fc4 (X11/20050929) |
Andi Kleen wrote: Yep - on x86-64. But the machine had 8GB of RAM, so if even that runs out of memory something is wrong imho. How large allocations would it need? -Andi 28 bytes per extent so in this case rather a lot. This could be compressed some at the cost of bit manipulation when looking up extents. I used to rack my brain about how to make xfs work with a partial in memory extent map. At least reorganizing the data structure to avoid the large memory requirement would be good. You probably didn't see it in the allocator because it was delaying between retrying the allocate. Steve |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: XFS unhappy with large holey loopback and syncs, Andi Kleen |
|---|---|
| Next by Date: | Re: XFS unhappy with large holey loopback and syncs, Andi Kleen |
| Previous by Thread: | Re: XFS unhappy with large holey loopback and syncs, Andi Kleen |
| Next by Thread: | Re: XFS unhappy with large holey loopback and syncs, Andi Kleen |
| Indexes: | [Date] [Thread] [Top] [All Lists] |