Results 1 to 2 of 2

Thread: [SOLVED] Proxy Problems - possible bug?

  1. #1
    Join Date
    Jan 2008
    Location
    Pretoria
    Posts
    133
    Rep Power
    7

    Default [SOLVED] Proxy Problems - possible bug?

    Hi I have a multi server setup and I am using a network load balancer to distribute across my 4 mail stores. I setup an external IP address on the load balancer and have configured the setup to use a virtualHostname and 4 virtualIPAddresses - each one being on a proxy. The idea is that you connect to FQDN and no matter which mailbox your account lives on, you always see FQDN in the url location bar.

    This has been working up until yesterday. What occasionally happens is that users go to the FQDN and log in, but then instead of the URL remaing as FQDN, it changes to mail1 or mail2 or whichever mail store their account resides.

    The proxy servers are not seperate servers, the service runs on the same devices as the mail stores.
    What is infuriating is that the problem is intermittent.

    What i have noticed in the nginx logs is:

    2008/09/10 15:27:48 [error] 24315#0: *545 upstream timed out (110: Connection timed out) while reading response header from upstream, client: 192.168.0.4, server: zimbra1-vm1.somedomain, URL: "/service/soap/NoOpRequest", upstream: "http://192.168.16.230:8080/service/soap/NoOpRequest", host: "mail1.somedomain", referrer: "http://mail1.somedomain/zimbra/"

    Now the IP address for the proxy on this server is 192.168.16.50 and it resides on the same machine as the 'upstream client' 192.168.16.230 so why it is getting a connection timed out is beyond me.

    How can I get more debug info out of nginx?

  2. #2
    Join Date
    Jan 2008
    Location
    Pretoria
    Posts
    133
    Rep Power
    7

    Default

    Ok - I had to eventually open a case regarding this and it turned out to be a configuration issue. There is not a lot of documenation surrounding standalone proxies and mailstores - so here how it was resolved.

    On all mailstores - disable the proxies with:

    /opt/zimbra/libexec/zmproxyinit -d -m -w -H `zmhostname`

    Then on all the proxies enable with :

    /opt/zimbra/libexec/zmproxyinit -e -w -H `zmhostname`

    The missing piece of documentation was the following - For all servers mailstores and proxies, the following must be configured:

    $ zmprov ms zimbraserver zimbraMailReferMode reverse-proxied

    Restart zmcontrol on all proxies and mailstores. Now we have 2 proxies load balanced and four mail stores working perfectly.

Similar Threads

  1. Bug with reverse proxy and wiki
    By FFL in forum Developers
    Replies: 2
    Last Post: 08-01-2008, 05:40 AM
  2. 5.0.6 is out!
    By mmorse in forum Announcements
    Replies: 0
    Last Post: 05-23-2008, 02:16 PM
  3. 5.0.4 is live!
    By mmorse in forum Announcements
    Replies: 0
    Last Post: 03-24-2008, 04:27 PM
  4. Possible Firefox 3 bug may cause Zimbra problems
    By iain in forum Administrators
    Replies: 5
    Last Post: 02-14-2008, 10:18 AM
  5. [updated]Perdition IMAP Proxy Remote Exploit Bug
    By jholder in forum Announcements
    Replies: 0
    Last Post: 11-06-2007, 07:36 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •