Page 1 of 2 12 LastLast
Results 1 to 10 of 12

Thread: Manual Fencing

  1. #1
    Join Date
    Mar 2006
    Posts
    6
    Rep Power
    9

    Default Manual Fencing

    I realize that manual fencing in RHEL 4 Cluster Suite is not intended for a production environment and is only intended for testing, but is it possible to install/run Zimbra Network Edition in such a manner?

    I tried installing it once. I had to do some manual configuration, since there are only the two power-fencing options in the cluster postinstall script. All the processes seemed to start OK up until "fenced" - which failed to start. In the message log it simply said: "fenced: startup failed".

    I could not access the web interface and when I ran clustat I got the following info:

    bajor not found
    realloc 1092
    endor not found
    realloc 1368
    Member Status: Quorate
    Member NameStatus
    ------ ---- ------
    bajor.domain.comOffline
    endor.domain.com Offline
    endorOnline, Estranged, rgmanager
    bajorOnline, Local, Estranged, rgmanager

    So I decided to start over. I uninstalled all the Zimbra components and am in the process of reinstalling. Last time, I attempted two active nodes. But this time, I'm trying one active and one standby node. But I'm wondering if I'm just wasting my time. Is this ever going to work without a physical power-fencing device? Thanks.

  2. #2
    Join Date
    Mar 2006
    Posts
    6
    Rep Power
    9

    Default Update

    Just wanted to post an update:

    Turns out that I couldn't get the nodes to come online because they could not contact my MTA (it wasn't running). As soon as I got it to run, the nodes showed as being online. However, I still cannot access the web interface because I cannot get tomcat to start. Whenever I run "zmcontrol start tomcat" I get:

    [zimbra@bajor etc]$ zmcontrol start tomcat
    Host bajor.domain.com
    Starting logger...Done.
    Starting mailbox...FAILED
    getService: antivirus
    getService: antispam
    getService: mta
    getService: sasl
    getService: webxml
    getService: perdition
    Can't open /opt/zimbra/tomcat/conf/server.xml.in: No such file or directory at /opt/zimbra/bin/zmmtaconfig line 545.
    sed: can't read /opt/zimbra/tomcat/conf/server.xml: No such file or directory
    mv: cannot move `/tmp/server.xml.1492' to `/opt/zimbra/tomcat/conf/server.xml': No such file or directory
    /opt/zimbra/bin/tomcat: line 124: /opt/zimbra/tomcat/logs/catalina.out: No such file or directory


    Starting snmp...Done.
    Starting spell...Done.

    Anybody know where I can find these files? Do I have to run a script to build them?

  3. #3
    Join Date
    Mar 2006
    Posts
    6
    Rep Power
    9

    Default

    Well, I found the files in one of the rpm archives, but I think that ultimately led me down the wrong path. I got several more errors. I ended up reinstalling ZCS and manually creating the directories /opt/zimbra/conf/tomcat and /opt/zimbra/log/tomcat (which were being linked to). That seemed to fix that error, but now I can't get the mailbox to stay running. I'll run:

    [zimbra@mail1 ~]$ zmcontrol start
    Host mail1.domain.net
    Starting logger...Done.
    Starting mailbox...Done.
    Starting snmp...Done.
    Starting spell...Done.
    [zimbra@mail1 ~]$ zmcontrol status
    Host mail1.domain.net
    loggerStopped
    mailboxStopped
    snmpRunning
    spellRunning

    [zimbra@mail1 ~]$

    Logger used to run, but I have a feeling that in the process of trying to get mailbox running, I broke logger. I'm currently searching other posts and probably start another thread, as I think that this is an unrelated issue.

    BTW: When I run clustat, I get the following:

    endor not found
    realloc 1092
    Member Status: Quorate
    Member NameStatus
    ------ ----------
    bajor.domain.comOnline, Local, rgmanager
    endor.domain.comOffline
    endorOnline, Estranged, rgmanager


    I'm still not sure what all that means. But I think there's progress. If anyone has gotten clustering to work, please let me know. Thanks.

  4. #4
    Join Date
    Aug 2005
    Location
    San Mateo, CA
    Posts
    4,789
    Rep Power
    19

    Default

    You are running the Network Edition and have the cluster scripts/pkg right? Just want to make sure.
    Looking for new beta users -> Co-Founder of Acompli. Previously worked at Zimbra (and Yahoo! & VMware) since 2005.

  5. #5
    Join Date
    Mar 2006
    Posts
    6
    Rep Power
    9

    Default

    Yep. I haven't run them yet. I was hoping to get everything configured right first. It is my understanding that you can get it running (as a trial version) before running the scripts. Correct?

  6. #6
    Join Date
    Aug 2005
    Location
    San Mateo, CA
    Posts
    4,789
    Rep Power
    19

    Default

    Quote Originally Posted by arcsnoy
    Yep. I haven't run them yet. I was hoping to get everything configured right first. It is my understanding that you can get it running (as a trial version) before running the scripts. Correct?
    I was pretty sure the scripts are what did the setup for you which is why I asked. I've never seen such odd problems with cluster for folks who follow the docs.
    Last edited by KevinH; 03-30-2006 at 08:38 AM.
    Looking for new beta users -> Co-Founder of Acompli. Previously worked at Zimbra (and Yahoo! & VMware) since 2005.

  7. #7
    Join Date
    Mar 2006
    Posts
    6
    Rep Power
    9

    Default

    Sorry. I completely misread your previous post. I thought you were referring to different scripts. Yes, I have downloaded the cluster packages and run the related clustering scripts. And I've followed the docs as closely as I could given the fact that I don't have as many servers/nodes as are in the example. I think, though, that my problem lies in some issue between my service names, node names, and plugging in the wrong thing in the Zimbra configuration, because the cluster comes online (or appears to, anyway) , but the services are having the problem coming/staying up. It's as if the mailbox nodes try to talk to the LDAP/MTA server and hit some configuration conflict and shut down. I'm probably going to reinstall ZCS again and/or increase the verbosity of the logging (because either the current logs aren't telling me what I want to know or I'm not looking at the right ones ).

  8. #8
    Join Date
    Jan 2007
    Location
    Minnesota
    Posts
    719
    Rep Power
    9

    Default Where are the cluster install.pl and postinstall.pl scripts?

    I'm sure they're under my nose somewhere, but I've been searching all day and can't find them anywhere.

  9. #9
    Join Date
    Mar 2006
    Location
    Beaucaire, France
    Posts
    2,322
    Rep Power
    13

    Default

    They're hidden 8)

    Once you have untarred the main ZCS-NE archive, you have to go in the "zcs/packages" directory and untar the cluster archive.

    Beware of the uninstall bug if you try to uninstall a clustered setup : the cluster RPM is not deleted thus next time you try to install the cluster, the cluster RPM is not setup (and you'll be missing files).
    Bug 17209 - ./install.sh -u does not delete cluster RPM

  10. #10
    Join Date
    Jan 2007
    Location
    Minnesota
    Posts
    719
    Rep Power
    9

    Default

    OK, I have a cluster running, sort of, but am having trouble figuring out how and when to actually make the cluster partitions (either qdisk or data). In the poorly linked RedHat documentation I have found circular references to "you need to make a simple change to lvm.conf for cluster locking to work," but nothing about what that change actually is.

    Also, is clvmd, for mediating LVM2 access in a cluster, an actual binary? I don't see it anywhere. I have the dlm, rgmanager, ccs, cman, and fence RPMs installed and running, though I don't have fencing correct (need to figure out a little bit more about IPMI on PowerEdge 2950s).
    Code:
    [root@mail1 ~]# pvs
      Skipping clustered volume group vgZimbra1Shared
      PV         VG      Fmt  Attr PSize  PFree
      /dev/md3   vgLocal lvm2 a-   62.19G 27.88G
    [root@mail1 ~]# vgchange -ae vgZimbra1Shared
      Skipping clustered volume group vgZimbra1Shared
    [root@mail1 ~]# clustat
    Member Status: Quorate
    
      Member Name                              Status
      ------ ----                              ------
      mail1.example.com                   Online, Local, rgmanager
      mail2.example.com                   Offline
    
      Service Name         Owner (Last)                   State
      ------- ----         ----- ------                   -----
      mail.example.com    mail1.example.com         started

Similar Threads

  1. Manual en español (o al menos en ingles)
    By nestor in forum Installation
    Replies: 4
    Last Post: 05-20-2008, 06:46 PM
  2. Manual edit of Postfix config files
    By ash in forum Administrators
    Replies: 1
    Last Post: 08-11-2006, 05:03 PM
  3. Manual postfix editing?
    By dhallweb in forum Administrators
    Replies: 6
    Last Post: 04-11-2006, 07:39 AM
  4. manual installation instruction
    By Brain in forum Installation
    Replies: 1
    Last Post: 10-03-2005, 02:22 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
  •