This post is more than 5 years old
6 Posts
0
1560
Maximum Celerra nfs.v3xfersize for old NS20 running version 5.6.40-3
Hello.
We are migrating off an old Celerra NS20 that is running version 5.6.40-3, and would like to improve the the efficiency of the migration by increasing the NFS rsize/wsize, which is controlled by the Celerra nfs.v3xfersize parameter. This parameter is currently set at 32kb, the default.
For version 5.6.40-3, can the nfs.v3xfersize parameter be increased? If so, to what maximum value? The Linux clients being used are capable of a 1MB rsize/wsize. The earliest documentation that I can find related to increasing nfs.v3xfersize is for version 6.0+, so increasing the parameter for version 5.6.40-3 my not be possible.
If the value can be increased, can it be done on-line, and what would be the artifacts? Ideally, existing NFS mounts would continue to use the previous setting, and new client mount requests could negotiate a higher value. Are there any additional settings that need to be added to the NFS share definition or export to allow for a larger rsize/wsize ... on this old version of DART
Thank you for your help.
Dave B.
Dave_Baril
6 Posts
1
July 19th, 2016 06:00
Hello Peter,
Thank you for your quick reply.
You provided two key pieces of information. 1) that the nfs.v3xfersize parameter could be modified in DART version 5.6 and before, and 2) that when the server_param "-info" option is used on an INDIVIDUAL parameter, it provides additional details that are not reported when information on multiple parameters are being requested.
I must admit, I missed the fact that additional details were provided when an individual parameter was queried with "-info". The server_param man page and its entry in the Celerra Parameters manual do not convey this information in
the description of the "-info" option, but if you read further down the man page or the Parameters manual, there is a clear example of the command output of "-info" which shows the level of detail.
For our Celerra NS20 running DART 5.6.40-3, here is the resulting output:
$ server_param server_2 -f nfs -i v3xfersize
server_2 :
name = v3xfersize
facility_name = nfs
default_value = 32768
current_value = 32768
configured_value =
user_action = none
change_effective = immediate
range = (8192,32768)
description = NFSv3 I/O transfer size
Unfortunately, this output shows that the maximum value is only 32kb in version 5.6, so increasing the parameter beyond 32kb is not possible.
The information also shows that the change can be done online with an immediate effect without rebooting the data mover.
Again, thank you for your help. It identified how to obtain the information that I was looking for.
Regards,
Dave B
Peter_EMC
674 Posts
1
July 19th, 2016 04:00
The v3xfersize could be changed in 5.6 and before.
I am unsure, if there was already a server_param command. But give it a try:
try a
server_param server_x -f nfs -i v3xfersize
It will list the range and if a reboot is required. I assume a DM reboot is needed, because it is still needed in newer versions.