RPC errors when backing up NT 4.0 machines

[mol1234]

Hello there, we are using a very old (4.26b) version of NetWorker.&nbsp;&nbsp;We are running this on Solaris 2.6 and have no problems at all backing up any of our Unix machines.&nbsp;&nbsp;However, we are running into RPC errors when it hits one of our NT machines.&nbsp;&nbsp;Specifically you get:<br><br>save, RPC error, unable to send<br>Send Chunk to MMD failed, Connection reset by peer<br><br>Stopping and restarting the client on the NT machine didn't seem to help, so we rebooted the machine (it had been up for a while anyway :&gt.&nbsp;&nbsp;It worked for about a day or so, now it's giving us the RPC error message again.<br><br>I can tell you too that that machine has both a 10-net IP address and a standard IP.&nbsp;&nbsp;It needs to be on our parent company's network as well as our own.<br><br>Any suggestions on how this could be fixed would be totally appreciated, as well...we need our backups on this machine!&nbsp;&nbsp;<br><br>thanks!<br><br>--alissa <br>

 

[ShawnL]

I don't know if this will help, but we found that on our network, if we are running SQL server and the machine isn't at full duplex, then we will get the MMD send chunk error.&nbsp;&nbsp;We haven't encountered it on anything but machines running SQL, but it still may apply to you.&nbsp;&nbsp;Try and hook both that client machine and your nw server into a switch and verify that their NICs are running at full and see if that makes any difference.

 

[Guest_imported]

I only came across this error once and eventually found it was a problem with the NIC on the client.Have you tried swapping it out for a new NIC or backing up over a different interface

 

[cfowler]

If you use the rpcinfo -p <clientname> command and you do not receive a response, use the nslookup command. If you can nslookup the name (nt server) and it returns an IP, but you get an &quot;unknown host&quot; when you nslookup IP, then you have to enter a PTR record into your DNS server. What this means is that the reverse lookup is not functioning on your network. I had a similar problem with database servers. I could backup and restore to the file servers, no problem, couldn't backup the database servers using the savepnpc command though. Once I had the PTR in place, databases flew!

Good luck,

CF

 

[cfowler]

RPC Errors, the worst! I have encountered rpc errors with the following issues:

Speed and duplexing of network cards, make certain they match those that are set on the switch. Legato reccommends 100 Mbps / Full Duplex. If server is set to auto / auto and switch is set to 100 Mbps / Full Duplex, this could cause an RPC error.

Indexes have become corrupt! Definitely can cause an RPC error. Run nsrck -F and then nsrim -X to correct the problem.

Alias within client's configuration may not be entered correctly. Easiest to solve. Simply add the missing alias within that client's configuration, stop the services on the client, restart the services on the client, you should be back in business. Also, take out the bad or inaccurate alias, if one or more exist.

Incorrect or modified hosts / DNS tables. Remember, if you are not responsible for the DNS table in your organization, check with the person who is, to see if any changes were made recently. I have encountered this one on a weekly basis, well almost weekly.

Make certain that all servers have a PTR in place on the network for those clients that shall be communicating via DNS. Use nslookup to ensure that the reverse lookup is operating correctly, if not, get it working.

The above are the only times I have encountered an RPC error.