E
Eddie
Hi,
Our company is facing a problem with an asp.net client connecting to a
web service. Basically we front-end it by a Cisco Content Smart Switch
load balancer which has a SonicWall attached to it to do hardware SSL.
The caller is in the same subnet/dmz as the webservice, but due to
business reasons we need it front ended by this hardware.
For about 99% of our transactions they are sucessful. The problem is
the last 1%. On these 1% of failures, the error message we get is:
"The underlying connection was closed: Could not establish secure
channel for SSL/TLS."
We've already brought this issue to Cisco, and they seem to have found
some strange connection reset problems. Cisco issued us a patch and
we've deployed them to our production environment, however the problem
still persists. I noticed that there are several people with the same
error string of "The underlying connection etc etc". I don't think
it's a certificate installation problem, as the web service works 99%
of the time.
The servers are currently running .net 1.1 sp1. I also confirmed that
the problem exists using .net 1.1, and .net 1.0sp2. They run Windows
2000 AS.
Is there any possible problems with the framework where if a
connection is reset by another device in the network that the
framework tries to use the previous connection it "knows" about,
rather than re-establish a new ssl connection? Once the problem
occurs, the subsequent request for the webservice is sucessful, and
then intermittently the problem occurs again.
Also, could there be a timeout where the established connection closes
on the client, and the framework wants to use the stale connection, at
that point giving the error message?
firewall
|
css css-+-+-sonicwall (Hardware SSL)
| / |
----- / -----
| | / | |
m m / w w
y y /|\ e e
s s / b b
e e /ssl s s
r r e e
v v r r
e e v v
r r i i
1 2 c c
e e
oversimplified diagram.... in this scenario the servers are in the
same dmz/subnet, but we do have clients connecting to the web service
from other dmzs.
Anybody else facing the same problem? Any fix?
Our company is facing a problem with an asp.net client connecting to a
web service. Basically we front-end it by a Cisco Content Smart Switch
load balancer which has a SonicWall attached to it to do hardware SSL.
The caller is in the same subnet/dmz as the webservice, but due to
business reasons we need it front ended by this hardware.
For about 99% of our transactions they are sucessful. The problem is
the last 1%. On these 1% of failures, the error message we get is:
"The underlying connection was closed: Could not establish secure
channel for SSL/TLS."
We've already brought this issue to Cisco, and they seem to have found
some strange connection reset problems. Cisco issued us a patch and
we've deployed them to our production environment, however the problem
still persists. I noticed that there are several people with the same
error string of "The underlying connection etc etc". I don't think
it's a certificate installation problem, as the web service works 99%
of the time.
The servers are currently running .net 1.1 sp1. I also confirmed that
the problem exists using .net 1.1, and .net 1.0sp2. They run Windows
2000 AS.
Is there any possible problems with the framework where if a
connection is reset by another device in the network that the
framework tries to use the previous connection it "knows" about,
rather than re-establish a new ssl connection? Once the problem
occurs, the subsequent request for the webservice is sucessful, and
then intermittently the problem occurs again.
Also, could there be a timeout where the established connection closes
on the client, and the framework wants to use the stale connection, at
that point giving the error message?
firewall
|
css css-+-+-sonicwall (Hardware SSL)
| / |
----- / -----
| | / | |
m m / w w
y y /|\ e e
s s / b b
e e /ssl s s
r r e e
v v r r
e e v v
r r i i
1 2 c c
e e
oversimplified diagram.... in this scenario the servers are in the
same dmz/subnet, but we do have clients connecting to the web service
from other dmzs.
Anybody else facing the same problem? Any fix?