Some suggestions..
If you download and run filemon (on both source and target machines) from
sysinternals.com and watch the file accesses as the batch file runs, they
will indicate if a failed file access failure is occurring and may help
pinpoint what kind of a failure is happening.
If the BIG file is copied temporarily locally to server_1 and the batch job
modified to copy from that "local source" this will eliminate the server_2
and if the problem still exists, tell you if the problem may lie in
something on the server_1 , anti virus s/w, bad disk sectors even.
Ditto the same on server_2, can the file be copied around on that pc without
problem to eliminate its disk failing.
If all works "locally" then really all you have is the network, do all BIG
file copies fail? we had some network devices which failed to "honour" the
tcp/IP MTU (Maximum transmission unit) value and when a PC filled a packet
with data over a certain byte count the packets were lost by the physical
network switch and the connection "hung"
Test this by (assuming its a tcp/ip connection) using the ping command with
an incrementing packet size and seeing how big the packets get before you
get packet loss. Some packet loss is to be expected but above a certain
value you may get 100% loss which indicates a configuration problem or
network switch malfunction.
do this from BOTH server_1 to Server_2 and server2 to Server_1
On Server_1, from a cmd prompt type
Ping -l 1024 Server_2 Repat this from the other server but change the
target name to Server_1
The -1 is a lower case L (LIMA) and the value is the byte size you want to
fill the packet with
Change the 1024 to 2048, repeat with 4096 8192 16384 etc
You will see it tell you it is pinging the (IPADDRESS) with xxx bytes of
data and hopefully a summary at the end with packet loss indicated (if any)
If this fails with any of these values I would be asking for the network to
be investigated, it is possible to restrict the size of a data packet that
is sent on a network card so the packet will never exceed the maximum you
have proven the network can accept but that MUST be seen as a temporary
issue.
Articles about PMTU and black hole discovery may be helpful if this is the
case.
Remember with most problems involving multiple variables (in yr case 2
servers, their o/s, anti virus s/w and a network), eliminate them one at a
time as best you can.
Forgive if it sounds patronising but in these forums no one knows the others
skill sets so I've pitched it at reasonable techy level is that OK?
Keith
Mark V said:
In said:
Hello,
There is a batch job with some copy commands in it, located on
server_1. It is run by an automatic task, and the copy command
"pulls" the file(s) from a share of server_2.
This works except with one big file (1.5G). The copy job hangs,
we dont know why yet.
I noticed: In the local computer management console, in the node
"Open Files" I see that the account which runs the copy job has
opened the file twice resp. there are two entries for READ on
the file.
Q: Is this "double opening" normal?
Anyone has an idea how to surround the problem?
[Win2K SP4]
Not immediately. But try using XCOPY instead of COPY for good
measure.
