UNSOLVED
Multiple VDM/CIFS servers and networking
I'm looking for some info on VDM/CIFS network configuration on a VNX.
In a particular configuration I have multiple VDMs on a VNX, and each VDM has a single CIFS server in it. Each CIFS server is on the same subnet. It looks something like:
- VDM1
- CIFS1 - 192.168.50.20/24
- VDM2
- CIFS2 - 192.168.50.21/24
- VDM3
- CIFS3 - 192.168.50.22/24
- VDM4
- CIFS4 - 192.168.50.22/24
And then we also have a replication IP for VNX Replicator in that subnet - 192.168.50.10/24.
These 5 interfaces all live on a single LACP trunk device.
I am seeing some very bizarre network behavior and odd netflows. Is there something about this configuration that won't work? I'm trying to rule out the VNX configuration as problematic before troubleshooting elsewhere. I've done similar configurations before with multiple CIFS servers on the same subnet/device, but never with the VDM separators in place.
Any info is appreciated.
1334
0
Posted June 27th, 2014 01:00
Hi mate,
For this issue, you could refer to the white paper http://www.emc.com/collateral/software/white-papers/h10741-vnx-data-movers-wp.pdf. I think the configuration is ok.
1339
0
Posted June 30th, 2014 07:00
Since you dont say what you actually see I can only guess
You might see the effects of packet reflect – see the VNX networking manual for details
1339
0
Posted June 30th, 2014 14:00
same class C subnet for all interfaces, packet reflect should not be an issue ?
809
0
Posted June 30th, 2014 17:00
if your switch ports are configured for tagging, you need to tag on the VNX (i always do) . If you have no plans to add another VLANs have them change ports to access/native VLAN.
1339
0
Posted June 30th, 2014 17:00
Sure it could
But without more info this is just speculation
1339
0
Posted June 30th, 2014 17:00
Sorry for the delayed response. Mostly I was just making sure that nothing was obviously wrong from a config perspective with respect to the VDMs.
The issues I am seeing make me suspect a networking or VLAN configuration issue. We are behind a trunk interface, yet if I tag the traffic on an interface (like I would normally need to do on a trunked interface), I lose connectivity.
Pinging from the interfaces to various things (including on the local subnet) is intermittently successful. Network ping from hosts to the interfaces is almost always successful.
Lots of errors in the DM logs about can't contact DNS and can't contact DC - even though oddly enough accessing CIFS shares seems to work. Somehow.
I know this is pretty vague, but as I said I'm pretty sure this is outside of the VNX config so I'm letting some other smarter folks on that side of the aisle take a whack at it.
Thanks for the input!
1339
0
Posted June 30th, 2014 17:00
how ? if you have all interfaces on 192.168.50.22/24 subnet, no tagging ...what other ways it can talk back to the client ?
809
0
Posted July 6th, 2014 02:00
When answering an off-net client it can matter whether to go through default route or reflect
733
0
Posted July 6th, 2014 06:00
Can you verify DNS settings primary and secondary DNS and port settings for your storage node. Sometimes they also create issues.
dynamox
11 Legend
•
20419 Posts
•
87439 Points
1334
0
Posted June 24th, 2014 18:00
don't see anything odd here, pretty simple if you ask me. No VLAN tagging, no static routes. VDMs are logical containers, have no impact on underlying network connectivity. What issues are you seeing ? Are you see these issue when replication is paused ?