If not, then it looks like the BE server's connection. I am assuming when you say 'the servers' you include the BE server in that and it's not at a remote DR site on a different subnet for example.ĭo you get the same timeout when you ping between servers that are not using teaming with a 1473 sized ping? If so, then it would appear that your switch is the culprit. So From my clearly infinite wisdom of MTU, I would think that the problem lies on the BE server since that is the only one that is having the issues. One of these servers (our Exchange if it matters" has 4 NICs that are in the same team through the HP utility.
I also tried from all the other servers to the BE server and they all gave the message of "Packet needs to be fragmented but DF set" as I was expecting. I did so with little help to performance. Symantec suggested I dissolve the team and remake it. The only difference I have been able to find is that the BE server has a pair of NICs that have been "teamed" together through an HP Utility. The servers are all physically in the same rack and all connected to the same physical switch. The network masks are correct and that is where I get confused. So according to the linked article my max packet size is 1500 as it should be (26 bit overhead). This sort of thing can happen if you have over zealous firewalling or your return routes are not right (particularly for ICMP - check those network masks on your servers).ġ472 works, 1473 just times out.
#BACKUP EXEC 2010 R3 DOWNLOAD HOW TO#
This explains how to troubleshoot black hole routers and there is a registry change you can make that might help. When this happens, lower the size and eventually you will find the maximum you can send without packet fragmentation. Eventually you will see what happens when you exceed the Maximum Segment Size (MSS). Try increasing a few hundred bytes at a time. Start with a size of 1400 - you should get replies. You can work out the MTU using ping with the no fragment flag set. If PMTU is not working, you might have a 'black hole router' on your network swallowing ICMP / PMTU discovery packets (or perhaps all ICMP) meaning your server doesn't know when to stop increasing the packet size, which causes retransmissions which kills your performance. The overhead can vary depending on various factors (such as if you a running VLANs etc). The MTU is often 1472 on a network set to 1500 frames the difference is overhead. Path MTU discovery tries to determine the largest MTU (Maximum Transmission Unit) possible on the network. If one of my production servers dies I can bring it up as a VM on the Backup Server while I get replacement hardware, run it live, then do a Headstart Restore to the new server while it is still running live, take a final incremental image of the VM and restore the changes, and only have maybe 5-10 minutes downtime for the cutover back to physical hardware. If my "Image Manager" server is the one that's boned, no big deal, I just rebuild it and get the backups running again, as nothing production is on the box.Įdit - Also, Headstart Restore is a very nice feature. This server runs Image Manager for collapsing images and validation, and also runs the Shadow Protect console to remotely manage the servers being backed up.
#BACKUP EXEC 2010 R3 DOWNLOAD SOFTWARE#
The key to making this software work good is to have a dedicated File Server to use as a target server for the backups. You are trying to use the software in like you would Backup Exec or something else.
0 kommentar(er)
