slowly network connection

[discretixit]

hi

i have netapp FAS 3020 and cisco 4560 they connect via 3 cable each 1 GB on etherchannel (no lacp) and when i copy a 10 G file it's take ~10 min ,the destination is win server 2003 also connect to 4560 with 1G,what the problem ?and what  is the best way to connect NepApp to ciscp switch?

[discretixit]

hi

i understand but the server name is different server name that i copy from and to ...

[danielpr]

Fine but over all there is a problem in the network. The result should be Ok for  "Performing transport layer diagnostics." In your case there are lot of Maximum receiving unit is not matching with the Netapp MTU size. Obviously there will be a packet delay in this case.

Thanks

Daniel

[discretixit]

hi

thanks,

maybe you know, where i increase the TCP receive buffer size for NFS on redhat ENT 4,i look in /proc/sys/sunrpc/ and didn't find the 7984 value.

[danielpr]

NFS mount command  will support the :rsize" "wsize" you can set these valuse for the mount points.

Thanks

Daniel

[discretixit]

hi

the current value on the client is rsize=32768,wsize=32768 i run netstat -i on server and client :

client :

Kernel Interface table
Iface       MTU Met    RX-OK RX-ERR RX-DRP RX-OVR    TX-OK TX-ERR TX-DRP TX-OVR Flg
bond0      1500   0 725185549      0      0      0 840612336      0      0      0 BMmRU
eth0       1500   0 695245994      0      0      0 503486433      0      0      0 BMsRU
eth1       1500   0 29939555      0      0      0 337125903      0      0      0 BMsRU
lo        16436   0 30525573      0      0      0 30525573      0      0      0 LRU

server :

Name   Mtu    Network      Address      Ipkts  Ierrs  Opkts  Oerrs  Collis  Queue
e0b    1500    *  *   370      0     87      0       0      0
lo     9188     127          localhost      134      0    134      0       0      0
vif0   1500   *        Filesrv        68m      0    74m      0       0      0

the mtu number is the same,what the problem ?

[danielpr]

MTU is fixed in system by the configuration. However during the TCP hand shake there will be MRU negotiation taken place, the lower the value negotiated then that will be fixed by the sender and the receiver. You case the client machine is saying lower MRU values because some buffer overflow and not capable of receiving 1500 size of packets. Hope you understood why the MRU is lower in this case.

I would recommend you to read the Data ONTAP 7.3.1 Online help Page

http://now.netapp.com/NOW/knowledge/docs/ontap/rel731/html/ontap/nag/networking/concept/c_oc_netw_iconfig_understanding_types.html

Thanks

Daniel

[discretixit]

hi

thanks,

i read https://now.netapp.com/Knowledgebase/solutionarea.asp?id=kb22946 article,according to this article which value in to change and to what ?

the current value are (on the client) :

net.ipv4.tcp_wmem = 4096        16384   131072

net.ipv4.tcp_rmem = 4096        87380   174760

net.core.rmem_max = 131071
net.core.wmem_max = 131071

[discretixit]

hi

another Question

how i check NFS and CIFS performances ? and how i know it's good performance?

[danielpr]

Please refer the following  TCP Tuning Guide..And also make sure that Flow control is full for sending and receiving TCP/IP packets in the CISCO switch.

http://fasterdata.es.net/TCP-tuning//

[discretixit]

hi

the show flowcontrol command give this result :

Port       Send FlowControl  Receive FlowControl  RxPause TxPause
           admin    oper     admin    oper

Gi3/43     on       on       desired  on          0       2825286

Gi3/45     on       on       desired  on          0       2636066

Gi3/47     on       on       desired  on          0       2524416

it's ok ?

[danielpr]

It doesnt look like admin enabled for receive packets. Please look into the following config guide.

http://www.cisco.com/en/US/docs/switches/lan/catalyst4000/7.5/configuration/guide/gigabit.html#wp1020443

Thanks

Daniel

[discretixit]

hi

thanks for your learning answers.

i did some more test,i notice that the problem occur in 3 Dell servers PowerEdge 700,800,1000 all of them connecting to 4506 switch.

i copy 1G file from the server to NetApp share and i take ~3 min(it's start with good copy rate and after 20 sec i see 2 min),in parallel i run sysstat -x 1 command on netapp and see that in the start of the copy the rate is good and after 5 sec the rate going down and after ~2min growing again and finish the copy process.

( please see attached file sysstat -x 1)

do you have any ideas?