CSM Probe Question

Hello,
We are currently running an http probe on the CSM which accepts a return code of 200 and 401 (because this application is single sign on and CSM does not have a user defined for it).
This application is having an issue where the web application is available and returning a 401 code, however in some cases the actual application instance is not available.
The only way we can see that is by looking at the HTTP stream:
HTTP/1.1 401 Unauthorized
WWW-Authenticate: NTLM
Content-Length: 0
Date: Thu, 05 Jul 2007 16:29:22 GMT
Server: Apache-Coyote/1.1
Connection: close
This connection close is the only value by which we can tell whether the application is working.
My question is there anyway we can use this value in the probe. I am quiet sure that its not possible but if anyone can confirm that will be great.
Thanks

Hi Gilles,
I was reading that CSM only supports on a HTTP Probes the request methods like "GET", "HEAD" and "URL", Not "POST".
It is possible to configure in a TCL script a HTTP Probe with "POST"?
I see in the manual (4.2(x) Release)that does not appears the generic tcl command "POST".
I will really appreciate your help.
Thanks
Hugo Rivas
Network Services
Data Center Triara

Similar Messages

Redundant CSM probes not working using OneArmedMode+PBR

In a redundant configuration: 2xCat6500 with one CSM each, using One Armed Mode when we use Policy Based Routing for return traffic the redundant CSM probes fail. If we use Source NAT instead everything works fine (both Active and Standby ok).
The problem is that we need to user PBR because the servers need to know the source IP and we want to assure a quick failover.

I'm pointing to the alias address. I didn't mention before but both C6500 have an IP interface configured in the Server Side VLAN and are using HSRP. I think the problem is related with that - when the redundant CSM sends the probe request, the response is routed to the active CSM. Maybe I need to define a specific PBR to the probes.

CSM HTTP-Probe question

We use two redundant CSM-Modules with version 3.1(6).
Now we have the need to check the content of a website in a probe configuration. The serveradmins shall have the possibility with a change of a value in the website bring up or down their server (of the view of the CSM). It's not sufficient to check the http returncode, I think we need the possibility to check the whole checksum of the website.
How can we do this ?
Thanks+Best Regards
HL

I don't think the CSM has this feature [like the CSS does have the hash function to detect if the content of a page changed].
However, you can create your own tcl script for the CSM to send an http request and look for a special string in the response from the server.
There are example of tcl script on the CSM software download page.
Regards,
Gilles.

CSM probe debugging

Hi,
i've tried to debug a non-scripted probe on my csm, but i can't see any output. What does the message "Health Monitor quiet mode: output error messages" mean, and how can i make those messages visible?
TIA, Stephan

Hi Gilles,
I was reading that CSM only supports on a HTTP Probes the request methods like "GET", "HEAD" and "URL", Not "POST".
It is possible to configure in a TCL script a HTTP Probe with "POST"?
I see in the manual (4.2(x) Release)that does not appears the generic tcl command "POST".
I will really appreciate your help.
Thanks
Hugo Rivas
Network Services
Data Center Triara

Business Object LB - Probe question

Has anyone ever worked with Business Object for load balancing ?
My main question concerns the probe to get all the content of a web page where I would need to follow all the links included in a probed web page to get all the content.
Otherwise the probe is useless. Anything at this point would help.
Thanks

Hi Nancy,
I can´t find it right now, but I do remember reading that there were different XML definition versions for SBO objects. In the version 1, table names where used. In the version 2, object names where used. Looks like the schema defines version 2, but SBO is using version 2.
I will check if I can found where I did read that. Try speaking with someone in SAP and about XML versions, perhaps they could help you with this info.
Hope helps,
Ibai Peñ

CSM general questions

I've many times worked on the CSS but this is my first on the CSM and have some functional questions regarding it that I cannot seem to find answers for in the CCO docs. I appreciate any/all input.
I am running a CSM in bridged (non-secure) mode with the client side being on the MSFC side and my questions are as follows:
1) Should I disable NAT for both the client and server side?
2) Should the real servers default GW be pointed back towards the CSM VIP?
3) Does the CSM replace the source IP address of the client (in the client request) with that of itself?
Thanks again.
Mike.

Mike,
For a vserver in a transparent caching setup, you would configure no nat client, no nat server, but for most other traffic, you will want nat server, no nat client.
If you are refeffing to "ip nat" under the interface in the MSFC config, remove that from the CSM vlans.
For bridged mode CSM, the real server gateway is the MSFC vlan address on the CSM client vlan of the CSM, not to the CSM VIP.
If you config nat client, you also must config a nat pool for the CSM. In this case, the client source IP is replaced with a nat pool address.
Peter

CSM configuration question

Customer wasnt current CSM config changed:
===================================
Customer's request is to make all requests go to real server 173.200.12.109 and if that server is down only then send all requests to backup server 173.200.12.110. But if server 173.200.12.109 comes back online then start sending all requests back to 173.200.12.109 and not use 173.200.12.110.
Questions:
==========
1. I think I have to put the two servers in VLAN 110 and not VLAN 12 and use the CSM in bridge mode by giving the same IP addresses of 173.200.12.8 for client VLAN 12 and server VLAN 110. Right?
2. There are two CAT 6500 with a CSM in each in Fault Tolerant Mode already configured and running and now need the above config changes added. Do I need to configure both CSM manually or if I configure one CSM will the other copy the config automatically.
3. I need to add VLAN 12 and VLAN 110 to Switch VLAN DATABASE, but not add VLAN 12 or VLAN 110 to the MSFC2 or should I only add VLAN 12 to MSFC2 and not VLAN 110.
4.Do new firewall rules need to be created for the two new server real IP addresses or VIP or non change required to the Firewall Rules.
4. What are some useful troubleshooting commands I can see if this doesn't work ? -
show module csm x connection detail, etc.

1 & 3) The servers will go into vlan 110. You need to create this vlan on the switch but there is not a routed interface for it.
The client vlan (vlan 12) will need a routed interface however it is not clear from your diagram where this should be. You will need to create vlan 12 on the switch as well. If you are putting a firewall in front of this vlan which it seems you are from your diagram then the routed interface for vlan 12 will be on the firewall not the MSFC. If you put it on the MSFC you will simply route round the firewall - probably not what you want.
2) Version 4.1(1) does not seem to support the command "hw-module csm 'slot no" standby config-sync" (worth checking tho). Looks like this was brought in on version 4.2. Without this command yes you will have to manually copy the config.
4) if you are not allowing through http ( assuming it is http) to the VIP already yes you will need a new firewall rule. That will allow application traffic. If you need direct access to the servers for management etc. then you will need to add in rules for those as well.
5) sh mod csm x reals
sh mod csm x vservers.
sh mod csm x ft
sh mod csm x conn
HTH

CSM Probes went down for 15 minutes

Hi all,
This morning all the probes went down on the csm module for exactly 15 mins and then came back up. There has been nothing else in the logs to indicate whey the went down. I have found a watchdog process which i think might have started back up the process for SLB. Has anyone ever come across this and what was the reason that the probes stayed down for exactly 15 mins.
Cheers
Kev

A possible workaround is to reset the card from the SUP console.
Try:
http://www.cisco.com/en/US/products/hw/switches/ps708/prod_release_note09186a00800fe64c.html

PIX loadbalancing woth CSM - probe problem

2 CSM/CATs on one side (FT)
2 CSM/CATS on other (also FT)
load balancing 2 PIX 535.
probing icmp pings only "direct" pix interface
the opposite interface will never answer to ping.
So switching off int in one pix make real FAILED on one side but other side still have working real and sends traffic to one leg PIX.
How to solve that ?

I thinking about that:
http://www.cisco.com/univercd/cc/td/doc/product/lan/cat6000/cfgnotes/csm_3_2/icn/fwldbal.htm#1037625
when Firewall 1 and Firewall 2 are pinged on directly connected interfaces then directly connected probe detect pix problem. But problem with whole PIX device is less typical than one of his interfaces down (ie. fiber patchcord unplug) than one (opposite/working) interface answers with ping and CSM sends traffic to that "real".
Great solution will be pinging opposite pix interface
but this isn't supported by PIX ASA. So i have tried
ping "any" ip behind pix which is currentl ip address of CSM VLAN.
When you had one PIX there is no a problem... but when you had two of them you need check both of them.. you defining static route:
ip_behind_pix VIA ip_pix_direct_int
Then thing not only about ECHO REQ but also on ECHO REPLY - there is no way to put static routing for those devices what active and standbys on both sides will detect pix interface errros...
There is no way to put REPLY on different gate than ECHO REQ...
Think of it drawing 6 icons, giving them 10 ip (2 for pix inside and outside, one for every CSM) adds
and then try set up static route that ping REQ and reply will go the same way. There is no such way...
IMHO 8-)

Csm probe problem

Hello,
I have the following problem after configuring/setting up tcp probe:
probe TCP tcp
interval 10
failed 30
vserver test:3389
virtual x.x.x.x tcp 3389
serverfarm test
inservice
serverfarm test
real a.a.a.a
inservice
real b.b.b.b
inservice
real c.c.c.c
inservice
probe TCP
Vserver shows o.o.s
serverfarm shows o.o.s probe failed for all servers
when I show probe, I get:
real vserver serverfarm policy status
a.a.a.a:3389 test:3389 test (default) OPERABLE
I have a separate VIP setup for each server without a probe and I can connect to them on port 3389, so I know
the application/servers are ok.
The csm is running ver 3.1(4)
Can you explain why the probe shows operable yet the serverfarm shows probe_failed?
Thanks,

if you remove the probe, is the vserver inservice ?
I would also recommend to test with a software more recent.
There was some probe issues in the past and they should be fixed with new releases.
Gilles.

LVOOP Custom Probe Question

Hello,
I just read up a lot about custom probes and want to use them in a class system I set up. I wanted to have the parent class be the only one that actually has a probe, as it can call a method vi to get some dynamic information with no problem. I set it up, and the general probe of the parent class works great in the parent class, but it doesn't get called when I probe any of the child classes. Is this how it should be working? I already set the probe in the parent class properties to use the custom probe, and I don't want to have to make a bunch of child probes. I wanted to have the probe only be a part of the class because if it is outside the class, the probe will load with labview, along with the children classes, which adds up to about 1000 VIs that get loaded, all for a probe I may not use unless I am using the class.
Any help would be appreciated.

Hi Britoa,
I found that this is a known bug. This was reported to R&D (# 46P7BRWS) for further investigation. Thanks you for reporting it though. We are always working to improve LabVIEW.
Nick R
NI

ACE HTTP Probe Question (HTTP Version)

Hello, I am wondering what HTTP version ACE uses when sending HTTP/HTTPS probes to a web server. I am currently running A2(3.3) if that has any bearing. TIA

Hi,
The default get request method is HTTP/1.1 when the ACE is sending the probe and i dont
think there is a way to change this default behavior.
Regards,
Siva

ACE http Probe Question

Hi there,
I have a probe and it keeps failing. I believe it might be something to do with the special character in the URL. The URL requires a port then a page after the server address:
probe http Server2_Probe
description Checks Server2 http is responding
ip address 10.x.x.x
interval 30
faildetect 4
passdetect interval 3
receive 5
request method get url :8100/gdesktop
expect status 200 200
Can someone point me in the right direction as to how I implement this probe if the :8100 port syntax is not allowed?
Thanks

Hi,
In the probe definition add
port 8100 - this changes the port used from the default of 80. Then remove :8100 from the url.
HTH
Cathy

CSM persistence question

i all,
in my balancing enviroment i need to sending multiple connection from different client to the same real server.
there is a solution for my problem?cookie?sticky url?
i'm very confused,it's my first configuration in this way.
best regards
Aghi

Aghi,
this is possible if there is something the CSM can use to identify the connections that have to go to the same server.
For example, if your application can set a cookie value like server=rs1, then we can create a config based on cookie sticky to achieve what you need.
If there is no such data to identify the server, then it doesn't seem to be possible.
If a human can't tell us how to distinguish a connection that needs to go to server1 from another one, a machine can't do job either.
Gilles.

Recommended CSM Probe Timers

Looks like 4 timer commands you can use on probes:
1) Interval - How often to normally Probe
2) Retries - How many consecutive normal interval probes have fail before marking server as failed
3) Failed - How often to probe after a server is failed to determine if it should be brought back online
4) Open - For TCP probes, how long to wait for a TCP socket to open.
What do you guys recommend for timer values.
Currently we're using 5-3-60-10
But, I'm wondering about the 10-second Open timer. 10 seconds for a TCP socket to open? That seems insanely long. I'm tempted to change it to 1 second.

The default interval is 120 seconds and it would take 3 probes to fail before it would bring down the server. You can lowering this and also using the command "fail action purge" so when a server fails it forces the user to disconnect.

CSM Probe Question

Similar Messages

Maybe you are looking for