Windows 7 computers have slow to access network resources and via RDP going to certain Server 2003 R2 servers but XP machines have no problems
wondering if issue might related network connection if change setting tcp stack, able files causing issue.
99% of servers accessed server 2003 server 2003 r2 server 2008 server 2008 r2 std/end fine, except small select few. trying figure out why , how happening. there few thousand people in company few thousand servers , desktops / laptops , trying see might happening them when people mysteriously can not server in timeley fashion.
i not willing blanket change company build image, nor authorized so, hoping might able assit me in trying figure out might changed on server side resolve other "99%" of servers in company 2003 2008 r2 not having issue.
microsoft windows 7 professional x64 sp1 running local administrator. if remotely connect microsoft windows server 2003 r2 standard edition sp2 x86 physical server ibm 7979-xxx xseries 3650 server broadcom bcm5708c netxtreme ii gige (ndis vbd client) card built it.
when users remotely connect remote desktop connection, have horribly slow connection, on 100mb network connection on company lan (local area network) on same physical switch in same physical office. if users on microsoft windows 7 go command prompt administrator, elevated permissions, , type in command below, able operate normally, without slowness! users on microsoft windows xp sp3 x86 not have problem. there issue windows explorer, if user on windows 7 tries go server via network share or folder share. to experience effect, must disconnect or log off of remote desktop connection (and connect server again), or log off or restart computer, windows explorer folders - accessing server in question.
netsh interface tcp set global autotuninglevel=disabled
if want go "normal" type in command:
netsh interface tcp set global autotuninglevel=normal
i have looked @ 2 ibm 7979-xxx servers, 1 of having problem , 1 not! one server 2003 r2 standard edition x64 edition sp2 (not having problem) , other x86 edition sp2!
i trying find out if there driver issue, 1 has version 4.1.3.0 (x64 edition) , 3.0.5.0 (x86 edition)
server 2003 r2 standard edition x86 edition sp2
driver version: 3.0.5.0
802.1p qos - disable
checksum offload - tx/rx tcp/ip checksum
ethernet@wirespeed - enable
flow control - auto
jump mtu - 1500
large send offload - enable
locally administred address - not present
receive side scaling - enable
speed & duplex - auto
i did see file , printer sharing microsoft networks properties shows maximize data throughput !network applications! radio button selected server optimization.
server 2003 r2 standard edition x64 edition sp2
driver version: 4.1.3.0
802.1p qos - disable
ethernet@wirespeed - enable
flow control - auto
ipv4 checksum offload - tx/rx enabled
ipv4 large send offload - enable
jump mtu - 1500
locally administred address - not present
receive side scaling - enable
speed & duplex - auto
i did see file , printer sharing microsoft networks properties shows maximize data throughput !file sharing! radio button selected server optimization.
i did go ibm updatexpress system pack installer , saw broadcom netxtreme ii driver windows shows bxvdbx.sys 6.2.80 new version , installed 3.0.7.0 , bxnd52x.sys 6.2.8.0 , 3.0.5.0. i thinking that might fix it, @ 2 differences above , seem see no difference in two.
at last, memory , processor fine in task manager , issue has been tested numerous users on , over, putting in command
netsh interface tcp set global autotuninglevel=disabled
to "enable" fast speed , changing
netsh interface tcp set global autotuninglevel=normal
when done, take advantage of features in this. not know does, autotuninglevel, know started microsoft windows vista , in microsoft windows 7, not in microsoft windows xp. i not want enable feature "disabled" microsoft windows 7 users, 99% of servers, not experiencing problem - except 2 microsoft windows server 2003 r2 standard edition sp2 x86 physical servers broadcom nic's ibm 7979-xxx.
is there possible setting on server causes or might reason happening? should update firmware or driver resolve problem? i curious server has this, broadcom bcm708c netxtreme ii gige (ndis vbd client) has driver version 5.2.17.0. more current 2 above, still have slow issues.
server 2003 r2 standard edition x86 edition sp2
driver version:5.2.17.0
802.1p qos - disable
ethernet@wirespeed - enable
flow control - auto
interrupt moderation - enable
ipv4 checksum offload - tx/rx enabled
ipv4 large send offload - enable
jump mtu - 1500
large send offload - enable
locally administred address - not present
number of rss queues - 8
receive buffers - 200
receive side scaling - enable
speed & duplex - 100 mb full (the switch 100mb switch)
transmit buffers - 750
hi,
the problem may cause nic settings. need schedule downtime problematic server, and go properties of nic driver , disable tcp checksum offload.
turn off network card offload functionality (disable tcp checksum offload):
right-click my computer > properties > hardware tab > device manager.
click network adapters > right-click broadcom bcm5708s netxtreme ii gige > properties.
click advanced tab > property: checksum offload > value: none.
advanced tab > property: large send offload > value: disable.
to resolve issue, can use individual network adapters instead of broadcom network adapter. , need contact nic vendor support further investigation. understanding highly appreciated.
a bug report of broadcom bcm5708c netxtreme ii gige listed in windows server 2008 feedback center.
broadcom bcm5708c netxtreme ii gige causes tcp checksum failures in vm (workaround)
best regards,
aiden
aiden cao
technet community support
Windows Server > Windows Server General Forum
Comments
Post a Comment