Your absolutely right to assume we have multiple smart connect zones. There are some nodes that are connected to 3 different subnets. While others are just connected to 2 different subnets. O, and each have dynamic pools. So are you saying that I should change that? I should not make each SC zone dynamic pools? I will provide the output for those command soon, but I will have to cut off the IP addresses.
Peter_Sero
4 Operator
•
1.2K Posts
0
June 11th, 2014 07:00
If you use several subnets with dynamic pools,
different nodes may end up with different "default routes",
and some nodes may not be able to reach back to your PCs (browser).
Either make sure your PC can be reached from the Isilon
through any subnet/route,
or have one static pool (providing one fixed IP per node)
with a static route assigned for that static pool.
# isi network list interface -w
is useful to get the full picture of the current IP assignments, plus
# isi_for_array -s 'netstat -r | grep ^default'
for the default routes in effect now.
chjatwork
2 Intern
•
356 Posts
0
June 11th, 2014 08:00
Peter_Sero,
Your absolutely right to assume we have multiple smart connect zones. There are some nodes that are connected to 3 different subnets. While others are just connected to 2 different subnets. O, and each have dynamic pools. So are you saying that I should change that? I should not make each SC zone dynamic pools? I will provide the output for those command soon, but I will have to cut off the IP addresses.
Peter_Sero
4 Operator
•
1.2K Posts
0
June 11th, 2014 09:00
Don't need to post it here, just check (with your network team) wether your PC can be
reached from each single Isilon node or not. As you found, the combinations of subnets
active on one given node become numerous... So you need to find a way
to guarantee that packets from any Isilon node can reach your PC
at any time. At least from any node where you want to use the webui.
Get the local network admin(s) with you in front of a big clean whiteboard
and try to actually visualize the current situation and the scenarios below,
and keep asking,
"Can it still happen that one node will be unable to send data to the PC?"
Either configure all involved networks (in the routers, not the Isilon) to
route relevant packets through to the admin PC network
--or--
add a subnet on the Isilon with a static pool, one IP per node(!)
(you may have this already), and a add static(!) route for that pool(!),
that permanently ensures a path to the admin PC network,
completely regardless of whatever the default route on a node might be.
That pool could also be "local" to the the admin PC network,
then no static route will be needed.
If still no luck, post pictures of the whiteboard