We seen to be having an issue whereby our database server
<div>gets to 90% or higher fragmentation. When it gets to this point </div><div>we would need to remove form production and defrag using the </div><div>xfs_fsr tool. The server does get a lot of writes and reads. Is</div>
<div>there something we can do to reduce the fragmentation or could</div><div>this be a result of hard disk tweaks we use or mount options?</div><div><br></div><div>here is some fo the tweaks we do:</div><div><br></div><div>
<div>/bin/echo "512" > /sys/block/sda/queue/read_ahead_kb</div><div>/bin/echo "10000" > /sys/block/sda/queue/nr_requests</div><div>/bin/echo "512" > /sys/block/sdb/queue/read_ahead_kb</div>
<div>/bin/echo "10000" > /sys/block/sdb/queue/nr_requests</div><div>/bin/echo "noop" > /sys/block/sda/queue/scheduler</div><div>/bin/echo "noop" > /sys/block/sdb/queue/scheduler</div>
</div><div><br></div><div><br></div><div>Adn here are the mount options on one of our servers:</div><div><br></div><div> xfs rw,noikeep,allocsize=256M,logbufs=8,sunit=128,swidth=2304 </div><div><br></div><div>the sunit and swidth vary on each server based on disk drives.</div>
<div><br></div><div>We do use LVM on the volume where the mysql data is stored</div><div>as we need this for snapshotting. Here is an example of a current state:</div><div><br></div><div><div>xfs_db -c frag -r /dev/mapper/vgmysql-lvmysql</div>
<div>actual 42586, ideal 3134, fragmentation factor 92.64%</div></div><div><br></div><div><br></div><div><br></div><div>Regards,</div><div>David Fuller</div><div><br></div>