| To: | Aaron Kulbe <akulbe@xxxxxxxxx> |
|---|---|
| Subject: | [xfs-masters] Re: PROBLEM: Page allocation failure |
| From: | Nathan Scott <nathans@xxxxxxx> |
| Date: | Tue, 20 Dec 2005 15:02:26 +1100 |
| Cc: | linux-kernel@xxxxxxxxxxxxxxx, xfs-masters@xxxxxxxxxxx, linux-xfs@xxxxxxxxxxx, hch@xxxxxxx |
| In-reply-to: | <4a10e1350512191355t5c0abe18k4269e0ebbabc147e@xxxxxxxxxxxxxx> |
| References: | <4a10e1350512191355t5c0abe18k4269e0ebbabc147e@xxxxxxxxxxxxxx> |
| Reply-to: | xfs-masters@xxxxxxxxxxx |
| Sender: | xfs-masters-bounce@xxxxxxxxxxx |
| User-agent: | Mutt/1.5.3i |
On Mon, Dec 19, 2005 at 03:55:56PM -0600, Aaron Kulbe wrote: > [1.] SUMMARY: During rsync and heavy I/O operations, we experienced > abnormally slow response times on the file server. > ... > [4.] Kernel version (from /proc/version): Linux version 2.6.8-24.14-bigsmp ( Try a recent kernel, both XFS and the VM have moved on in leaps and bounds since 2.6.8 timeframe. > I tried Christoph Hellwig's suggestion from > http://oss.sgi.com/bugzilla/show_bug.cgi?id=345 of increasing the value of > /proc/sys/vm/min_free_kbytes. I set the value to 1024. The results were > the same. No change in performance, and errors persisted. There were older kernel versions where that parameter was not being honoured particularly well, I can't recall anymore if it was 2.6.8 or not though - it may well have been. cheers. -- Nathan |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | [xfs-masters] PROBLEM: Page allocation failures, Aaron Kulbe |
|---|---|
| Next by Date: | [xfs-masters] Re: Recovery from Accidental Hot Swap, Russell Cattelan |
| Previous by Thread: | [xfs-masters] PROBLEM: Page allocation failure, Aaron Kulbe |
| Next by Thread: | [xfs-masters] Recovery from Accidental Hot Swap, Steve Strobel |
| Indexes: | [Date] [Thread] [Top] [All Lists] |