[sam-qfs-discuss] Changing lease times on the fly

[Dean Roehrich]

On Mon, Dec 17, 2007 at 04:07:47PM -0600, jtw wrote:
> Thanks for your help,
> 
> Maybe in the documentation there could be a note that using "ms" devices on 
> large numbers of luns is not recommended due to cpu spikes.  We have added 
> another system with a similar number of users on ma (mm,mr) devices and have 
> seen a huge improvement.
> 
> It seems like syncing the metadata on "ms" was taking too long and causing 
> everything else to be delayed which then spikes the cpu.

I've had time to study the samtraces now and it does appear that we're
spending a lot of time merging "small blocks".

Here are two complete Syncs you caught in your samtraces:


12/13 15:23:53.195 15=30001d0d600 1-030000259000* Sync:     flag=1,lock=0
  -- lots of inodes being sync'd, then the last one:
12/13 15:23:53.386 15=30001d0d600 1-0600158e25c0  Contig:   ...
  -- dead air, probably in sam_merge_blocks()
12/13 15:23:53.642 15=30001d0d600 1-030000259000* Sync-     flag=1,lock=0

12/13 15:23:53.642 15=30001d0d600 2-030000259840* Sync:     flag=1,lock=0
  -- lots of inodes being sync'd, then the last one:
12/13 15:23:53.791 15=30001d0d600 2-06001707c500  Contig:   ...
  -- dead air, probably in sam_merge_blocks()
12/13 15:23:54.427 15=30001d0d600 2-030000259840* Sync-     flag=1,lock=0


Pretty sad.

We have made some improvements in this area but the best of that stuff hasn't
been queued up for 4.6 yet, and it's not clear those improvements are
sufficient, anwyay.

And yes, now I agree that ms/md is the cause of your headaches.

Dean