[Resolved] Ping URL/Documentation

Discussion in 'General' started by GOT, Oct 21, 2014.

Thread Status:
Not open for further replies.
  1. GOT

    GOT Well-Known Member

    Is there a reference document for lslb? I went to support-documentation and it seems there is only docs for lsws, not lslb. Also the wiki doesn't have much info either.

    Right now specifically, I would like to know how the ping URL functions. Here is the situation:

    We have nginx running on port 80 connecting to a python application that runs on port 5015. We had a situation where nginx was up and responding on port 80, but the backend python application was down. The load balancer didn't detect this and take server 1 out of the rotation presumably because nginx was up and responding even though it was throwing a bad gateway error.

    So, I presume I need to put a ping url in that checks the python application directly (though then this would not detect and nginx outage would it?)

    Its also confusing because while there are two workers in the group, there is only one ping url, so how does it know which server is down?
     
  2. GOT

    GOT Well-Known Member

    I tried simply putting a / in for the ping url, however, it did not take a server out of the cluster even when it was throwing 502 bad gateway errors.
     
  3. mistwang

    mistwang LiteSpeed Staff

    The LSLB documentation should come with the LSLB package, you can access it through the webAdmin console.
    The ping URL, you need to put the full url
    Code:
    http://domain.name/path/
     
    Last edited by a moderator: Apr 21, 2017
  4. GOT

    GOT Well-Known Member

    Yeah, the docs are not clear either.

    To reiterate my problem, I don't really understand the mechanism. One url cannot monitor both servers in the cluster. Or, if it does, such as with the public facing url, how does it determine WHICH server is offline.

    Can you elaborate please?
     
  5. mistwang

    mistwang LiteSpeed Staff

    LB sends the request with a public facing URL to both backend server and expects 200 responses, if the response is not 200, LB mark the corresponding backend as off-line.
     
  6. GOT

    GOT Well-Known Member

    OK, I guess that should be obvious. It did work fine, thank you.
     
Thread Status:
Not open for further replies.

Share This Page