11-17-2010 08:42 AM
I am a CSS newbie, these switches set up before me starting here and no one knows how they work or are configured. Here is a sanitized configuration. The traffic is only going through "interim5". ANY help will be GREATLY appreciated as we are in a bind.
CSS11501# sh run
!Generated on 11/17/2010 08:42:37
!Active version: sg0810503
configure
!*************************** GLOBAL ***************************
ip redundancy
acl enable
app
app session zzz.zzz.1.2
ftp-record TERMINAL xxx.xxx.11.171 USER des-password fffffffffffffffffffffff
ftp-record MYTERMINAL xxx.xxx.11.177 CSSUSER des-password ffffffffffffffffffff
ip route 0.0.0.0 0.0.0.0 xxx.xxx.11.170 1
!************************* INTERFACE *************************
interface e1
bridge vlan 2
interface e4
redundancy-phy
interface e8
bridge vlan 10
!************************** CIRCUIT **************************
circuit VLAN1
redundancy
ip address xxx.xxx.11.173 255.255.255.248
circuit VLAN2
ip address zzz.zzz.1.1 255.255.255.0
redundancy-protocol
circuit VLAN10
redundancy
ip address xxx.xxx.11.179 255.255.255.248
!************************** SERVICE **************************
service interim4
keepalive type tcp
keepalive port 80
ip address xxx.xxx.11.180
active
service interim5
keepalive type tcp
keepalive port 80
ip address xxx.xxx.11.181
active
!*************************** OWNER ***************************
owner OFFICE
content MIDDLE
vip address xxx.xxx.11.169
advanced-balance ssl
add service interim4
add service interim5
active
!**************************** ACL ****************************
acl 1
clause 1 permit tcp any destination xxx.xxx.11.169 eq 80
clause 2 permit any xxx.xxx.11.169 destination any
clause 3 permit any xxx.xxx.11.177 destination any
clause 4 permit any xxx.xxx.11.178 destination any
clause 5 permit any xxx.xxx.11.180 destination any
clause 6 permit any xxx.xxx.11.181 destination any
apply circuit-(VLAN1)
acl 2
clause 1 permit any any destination any
apply circuit-(VLAN2)
acl 10
clause 1 permit any any destination any
apply circuit-(VLAN10)
CSS11501#
11-17-2010 09:02 AM
The commands "show service summary" and "show rule OFFICE MIDDLE services" should give us some insight as to the state of the services and the rule to which they are assigned.
Please post the output of those commands.
BTW it appears you have a high availability pair. "show redundancy" should show you the status of that setup. Be advised that any configuration changes must be manually synchronized: "script play commit_redundancy "standy_unit_redundancy_interface_ip"" is the command you need to use (substituing the actual partner's IP but keeping it in quotes).
Also, if you're not running the most current system software, loading it is a good idea (assuming you have support). 8.20.4.02 is the latest release and it incorporates a number of bug fixes.
11-17-2010 10:23 AM
Thanks for taking a look, here is the output of those commands
CSS11501# sh service summary
Service Name State Conn Weight Avg State
Load Transitions
interim4 Alive 10 1 2 0
interim5 Alive 37 1 2 0
CSS11501# show rule OFFICE MIDDLE
Name: MIDDLE Owner: OFFICE
State: Active Type: HTTP
Balance: Round Robin Failover: N/A
Persistence: Enabled Param-Bypass: Disabled
Session Redundancy: Disabled
IP Redundancy: Not Redundant
L3: xxx.xxx.11.169
L4: Any/Any
Url:
Redirect: ""
TCP RST client if service unreachable: Disabled
Rule Services & Weights:
1: interim4-Alive, S-1
2: interim5-Alive, S-1
CSS11501#
11-17-2010 12:24 PM
Your services are showing as active but the application running on them may not be working correctly, which may be giving you the indication of the CSS not load balancing. The server can respond positively to TCP port 80 even if the application is down.
Dave
11-17-2010 10:33 AM
A little more back story, one of the load balanced systems went down for a while, not sure if there is something thats needs to be done to reintroduce it back into the load balancing, but I suppose that would defeat the purpose of load balancing fault tolerance.
11-17-2010 11:57 AM
From the command output, both your services appear to be active and the rule is seeing them as such.
Can you share what indicator is telling you that you are only getting traffic on one of the real servers? The CSS appears to think that traffic is hitting both services. You can confirm this by successively entering the "sh rule OFFICE MIDDLE services" command and watching the hits increment. (Note the "services" parameter at the end of that command.)
Looking at your configuration, I am not sure why they put "advanced-balance ssl" in the rule. As a port 80 VIP I would assume you are running http, not https. According to the command reference, the usage of that command is for:
Enables the content rule to stick the client to the server
based on the Secure Sockets Layer (SSL) version 3
session ID assigned by the server. The application type
must be SSL for the content rule. You must specify a
port in the content rule to use this keyword. The CSS
will then spoof the connection.
11-17-2010 12:22 PM
CSS11501#
CSS11501# show rule OFFICE MIDDLE services
Rule Services:
Local Load Threshold: 254
PrimarySorryServer: None
SecondSorryServer: None
VIP Ping Response Decision: Local Services Only
Name: Hits: Wgt: State: Ld: KAlive: Conn: DNS:
----- ----- ----- ------ --- ------- ----- ----
interim4 125,550,747 S-1 Alive 2 TCP-80 7 0
interim5 298,278,777 S-1 Alive 2 TCP-80 32 0
Location Cookie Services:
Name: Hits: State: Ld: KAlive: Conn:
----- ----- ------ --- ------- -----
Last Clearing of Stats Counters: 11/16/2009 16:12:07
CSS11501#
CSS11501#
11-17-2010 01:25 PM
drocks,
If you are concerned that the hit counts to the servers are not even, you could use the "balance leastconn" command on the content rule to help even things out.
interim4 125,550,747 S-1 Alive 2 TCP-80 7 0
interim5 298,278,777 S-1 Alive 2 TCP-80 32 0
Dave
11-17-2010 05:34 PM
Actually since those hit counts are cumulative (from the last CSS reboot, last zero or rule creation, as applicable) and the original poster mentioned the one server had been down for a while, that would be normal. That's why I suggested doing a reset of the counters and watching from that point forward. (Alternatively one could run the command repeatedly and expect to see increments in roughly equal amounts.)
11-18-2010 06:19 AM
I agree.
Dave
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide