2

We're having an issue with our linux systems connecting to our netapp via NFS. The read/write IOPS are only around 1000/s but the GETATTR calls are almost 10,000!

I found this bug but it says, it was fixed in kernel-2.6.18-92.el5 and we are on 2.6.18-274.el5.

We are mounting these with noatime.

Any ideas on a fix?

Diamond
  • 9,001
  • 3
  • 24
  • 38
DEVDEVDEVDEV
  • 21
  • 1
  • 2

2 Answers2

1

Disable access time on the NetApp volume, forget the client 'noatime' setting. You can run your IO load under strace and see what is being retrieved read from the export if needed (ie ratio of meta data to data IO).

pfo
  • 5,700
  • 24
  • 36
0

This is to improve the NFS performance as netapp is suggesting this file edit / config modify or VI /etc/sysctl.conf ( add the following 2 lines and reboot sometime in the near future )

#Set the number of outstanding RPC requests that can be present on a single NFS connection sunrpc.tcp_max_slot_table_entries=128

XIAJZ
  • 1