CSS Keepalive error messages

mdavis · ‎07-13-2005

I am running a pretty simplistic HTTP/uri keeplalive to a group a IIS servers and I am wondering about the meaning of the keepalive errors reported when issuing the show keepalive command. The keepalive error messages cycle from "General failure", "Not reachable", "Read failure". Bottom line the service is down, but it seems that these messages have more meaning and understanding the differences could help in troubleshooting.

Also if anybody knows of a command that would show more detail than the show keepalive commands. I am hoping to find something that give more information than just that the keepalive is dead or alive.

This is on a CSS11503 with 7.20 code.

Gilles Dufour · ‎07-13-2005

what method do you use ? GET or the default HEAD ?

With a HEAD, the CSS expect a response code of 200 and that's it.

With a GET, we expect a response code of 200 and will also hash the content of the page and compare it with the previous hash to verify that the content did not change.

Usually to solve this kind of issue, the best solution is to capture a sniffer trace on the server.

www.ethereal.com provides free software for all kind of platforms making it very easy to capture such trace.

I believe not reachalbe means no response to SYN.

Read failure means we did not get a response to the http request and general failre means no response code 200.

What you can try to do is the following command from debug mode [enter debug with the command llama].

CSS11503(debug)# icp probe ser linux1 "/"

Probing 192.168.30.27:80(/) KeepAlive probe..

IP Address: 192.168.30.27

Port: 80

URL: /

HTTP Version: 1.1

Server Model: Apache/2.0.46 (Red Hat)

Server Date: Wed, 13 Jul 2005 22:04:18 GMT

HEAD Response: 200 OK

HEAD Support: Yes

Persistence: Yes

Keep-Alive: Yes

Request Depth: 100

TBR: 44

Hash: 09e12709c8ee93589d4f44bd8e47099b

Connect Time: 1 ms

Rqst/Rsp Time: 365 ms

Pipeline: No

SSL: No

Regards,

Gilles.

Please rate this answer.

tevfik · ‎12-14-2005

Hi,

I have a same problem.

I set a named keepalive config. for port 80 and 81.

80 is ok but named keepalive for port 81 is not work.

Here is the config and debug:

!************************* KEEPALIVE *************************

keepalive SPDA_Transcoding_80

port 80

ip address 10.X.Y7.49

uri "/KeepAlive.bin"

type http

active

keepalive SPDA_Transcoding_81

port 81

ip address 10.X.Y.49

uri "/KeepAlive.bin"

type http

active

----------------------

Name: SPDA_Transcoding_80 Index: 1 State: Alive

Description:

Address: 10.X.Y.49 Port: 80

Type: HTTP:HEAD:/KeepAlive.bin

Encryption: Disabled

Frequency: 5

Max Failures: 3

Retry Frequency: 5

Dependent Services:

SPDA_AnTrTS01_port80

Name: SPDA_Transcoding_81 Index: 2 State: Down

Description:

Address: 10.X.Y.49 Port: 81

Type: HTTP:HEAD:/KeepAlive.bin

Keepalive Error: Not reachable failure

Encryption: Disabled

Frequency: 5

Max Failures: 3

Retry Frequency: 5

Dependent Services:

SPDA_AnTrTS01_port81

------------------

A_LB_MASTER# sh keepalive-summary

Keepalives:

SPDA_Transcoding_80 State: Alive 10.X.Y.49

SPDA_Transcoding_81 State: Down 10.X.Y.49

------------------------------------

A_LB_MASTER(debug)# icp probe ser SPDA_AnTrTS01_port81 "/"

Probing 10.X.Y.49:81(-) KeepAlive probe..

IP Address: 10.X.Y.49

Port: 81

URL: /

HTTP Version: 1.1

Server Model: Apache/1.3.33 (Unix) HTTPGW/1.0

Server Date: Wed, 14 Dec 2005 12:12:00 GMT

HEAD Response: 200 OK

HEAD Support: Yes

Persistence: Yes

Keep-Alive: Yes

Request Depth: 100

TBR: 15

Connect Time: 1 ms

Rqst/Rsp Time: 1 ms

Pipeline: No

SSL: No

----

As a notice, both services (two service) run on the same server that IP address 10.X.Y.49

Can anybody point my problem?

Gilles Dufour · ‎12-15-2005

As I said in one your many posts about the same question is that you need to capture a sniffer trace on the server to at least confirm that the CSS is sending requests and that the server is responding.

Then, if you did not already installed the very latest version, you may want to do so based on the following bug

CSCei81533

Class A keepalives get stuck into a DOWN state

Thanks,

Gilles

georges.merhej · ‎12-11-2007

Hello Gilles,

I'm facing a similar issue.

I have 2 IIS web servers running on windows platforms and 2 apache web servers running on solaris paltforms.

The keepalive icmp type works well for all the servers. When changed to HTTP keepalive, the IIS servers go down with error: General failure; the servers on solaris work great with HTTP keepalive. I took a capture with ethereal attached here; it shows that there is checksum error in the HTTP head!

What do u think the problem is ?

The version is 8.10 not affected with the bug CSCei81533.

Thanks,

Georges

Syed Iftekhar Ahmed · ‎12-11-2007

It looks as if your web servers are sending 302 redirects in response to your probes.

Both HTTP HEAD and HTTP GET probes expect "200 ok" response from web servers before marking them active.

You can ask application people to provide you a static page for probing purposes.This way you will get 200 response from your probes.

Syed Iftekhar Ahmed

georges.merhej · ‎12-12-2007

Hello Syed,

That's the exact proble. But I configured the workaround on the CSS itself with the command "keepalive http-rspcode 302".

Thank you.