The Community Forums

Interact with an entire community of cPanel & WHM users!
  1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

[FreeBSD] Do NOT restart Bind !

Discussion in 'Bind / DNS / Nameserver Issues' started by eos1, Sep 27, 2004.

  1. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    [FreeBSD] Do NOT restart Bind & Exim!

    Hi FreeBSD users,

    Do NOT restart Bind !

    I just restarted the Bind, everything was broken!!!
    I know nightly upcp was trying to upgrade Bind 9.2.3 to 9.3.

    Now, I can't SSH!!!

    This is VERY VERY BAD.

    PS: This box was doing VERY well.

    Thanks.
     
    #1 eos1, Sep 27, 2004
    Last edited: Oct 12, 2004
  2. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    Reboot, upcp to Current version fixed it.
     
  3. tprice42

    tprice42 Active Member

    Joined:
    Sep 10, 2004
    Messages:
    37
    Likes Received:
    0
    Trophy Points:
    6
    I've also been auto-upgraded to bind 9.3 and found that i suddenly get this problem when editing DNS records:

    Bind reloading on cpanelhostxxx using rndc zone: [xxxxx.co.nz]
    Error reloading bind on cpanelhostxxx: zone reload queued

    This error message doesn't appear to effect the opperation of bind as it reloads normally but the error message i could see raising some questions from my resellers.


    On a more important note i am finding that when the server boots it starts bind using /etc/namedb/named.conf, after adding an account bind is reloaded using /etc/named.conf. Both configuration files are identical in their structure however bind using "/etc/named.conf" doesn't seem to work and i have to restart bind from WHM before it acknowledges the hosting of the domain.

    In the past i've simply removed the /etc/named.conf file and symlinked the other file to that location but that was for a different problem and i'm not sure it will cure the fault. I'm sure that the cpanel developers will make some changes to accomodate bind 9.3 but in the meantime is anyone else having this problem? I'm running WHM 9.7.2 cPanel 9.7.7-R16
    FreeBSD 5.2.1-RELEASE i386
     
  4. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    I found out that nightly upcp broke /var/named/domain.com.db.
    I just manually corrected now...

    What's going on, Nick?

    This is really too much problems about Bind and Exim.

    Looks like this time, nightly upcp was trying to upgrade the Bind.
    some of them got upgraded, however it broke /var/named/domain.com.db.
    Boxes that didn't upgrade ---> check what's the bind in your system...if there is no bind 9.X ---> this will be the problem that I just got. (you can't even SSH).

    "Can't SSH" was already happened to 2 FreeBSD machines!
    One of them got totally destroyed including Kernel. This happened by only "upcp". Adam(cPanel) is looking into it, but I have no idea if he can fix...

    This is CRITICAL problem.
     
  5. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    Important

    IMPORTANT

    Check your Bind version at WHM: Service Status.

    If you see "bind ()", do NOT restart the Bind!!!
    do upcp first!

    or else, you'll be like me... :D

    If you see "bind (9.3.0)", then try restart the Bind. and check any errors.

    Just a note...


    PS: These days, I'm a chicken to reboot or restart services... :(
     
    #5 eos1, Sep 27, 2004
    Last edited: Sep 27, 2004
  6. easyhoster1

    easyhoster1 Well-Known Member

    Joined:
    Sep 25, 2003
    Messages:
    659
    Likes Received:
    0
    Trophy Points:
    16
    Yes, this has been going on for awhile now.

    Even running Bind 8 the zones are dropped into /etc/namedb/named.conf

    In the named.conf file, you will see the path to the db file has changed from
    /var/named to /etc/namedb. To fix this so the path to the zones in /var/named can
    still be reached we symlinked the named.conf

    copy your named.conf to /etc/namedb
    remove the named.conf from /etc and
    to symlink, run

    ln -s /etc/namedb/named.conf /etc/named.conf

    However, restarting Bind from WHM still creates a error and there is nothing in the logs
    to help trouble shoot why Bind fails. A reboot will be needed. Adding and removing domains
    is no longer a issue with the symlink.

    Hope this helps some :rolleyes:
     
    #6 easyhoster1, Sep 28, 2004
    Last edited: Sep 28, 2004
  7. ttiger

    ttiger Member

    Joined:
    Sep 1, 2004
    Messages:
    11
    Likes Received:
    0
    Trophy Points:
    1
    I have this problem also. When is this going to get fixed??????
     
  8. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    confusing...

    was:
    NOW:
    My reseller thought there is something wrong with rndc...it's confusing...
     
  9. Hantai

    Hantai Member

    Joined:
    Feb 9, 2004
    Messages:
    18
    Likes Received:
    0
    Trophy Points:
    1
    Location:
    London, UK
    There is indeed a problem with rndc as far as I can tell, please excuse me if I'm totally offbase as this isn't my forte, I tried various bind restart procedures and eventually got an error stating ndc couldn't be found, looking at ndc it's a link to rndc, and rndc in turn is quite simply missing, that can't be right can it? In any event I'm stuck with DNS down and am indeed showing Bind() in service status, what's the definite fix (even if just for temporary results) here? I'm about to try upcp again and see if that makes a difference...

    Addendum: it did make a difference, bind is now showing 9.3.0 and starting normally again. The missing rndc file is now present and accounted for, so I'm guessing there's a definite link of some sort there. Someone seriously needs to be shot for this tho :P.
     
    #9 Hantai, Oct 8, 2004
    Last edited: Oct 8, 2004
  10. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    Adam fixed it yesterday!!!:eek:
    Problems were my customer's scripts and .htaccess setup. I couldn't believe how easy to destroy cPanel...
    Thanks soooooooo much, Adam.
     
  11. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    [FreeBSD] upcp to 9.9.1-EDGE_9 messed EXIM AGAIN.
    downgrading to 9.9.1-CURRENT_5 fixed it.
    Just a note...
     
  12. Hantai

    Hantai Member

    Joined:
    Feb 9, 2004
    Messages:
    18
    Likes Received:
    0
    Trophy Points:
    1
    Location:
    London, UK
    Keep in mind that by definition Edge builds have a strong potential to mess things out, stick to Current or better for production servers unless you have a desperate need for something present only in an edge build.
     
  13. easyhoster1

    easyhoster1 Well-Known Member

    Joined:
    Sep 25, 2003
    Messages:
    659
    Likes Received:
    0
    Trophy Points:
    16

    Had the exact same error. Tried to downgrade to from Current to Release version did not fix it. We found that the /usr/local/lib/libcrypto.so.3 file was missing. We had to copy it over from another machine and this fixed the Exim problem.
     
  14. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    [FreeBSD] exim was dead.

    Hi all,

    One of box running exim (4.42+27), just now, exim was dead.
    running /scripts/exim4 fixed.
    exim (4.42+27) ---> exim-4.43+28

    Looks like related with spamd...

    PS: Others running exim (4.42+27) are OK so far, but might need to upgrade exim...

    Problem here is:
    -You don't receive any alerts.
    -cpsrvd failed to report that exim was not working.

    To find this issue:
    -Your customers might be missed... <--- I got this one :D
    -restarting exim


    Just a note for FreeBSD users.

    Thanks.

    Ca$h
     
    #14 eos1, Oct 12, 2004
    Last edited: Oct 12, 2004
  15. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    :eek: Another Box's Exim failed...after nightly upcp.

    Just a note...
     
  16. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    cp -R /usr/ports/dns/bind9/

    This issue is BACK!

    Copy/prepare bind 9.3.0

    nightly upcp will upgrade bind 9.3.1
    before this, copy /usr/ports/dns/bind9.

    Contacted cPanel, but looks like no fix for it...
     
  17. Hantai

    Hantai Member

    Joined:
    Feb 9, 2004
    Messages:
    18
    Likes Received:
    0
    Trophy Points:
    1
    Location:
    London, UK
    Any word on a fix for this? Rebooting and applying upcp as per the last time this happened doesn't appear to be effective in this case.
     
  18. jawah

    jawah Registered

    Joined:
    Nov 11, 2003
    Messages:
    3
    Likes Received:
    0
    Trophy Points:
    1
    I found a temporary fix for it: New Bind installed itself into /usr/local/sbin instead of /usr/sbin so I made a symlink like this

    ln -s /usr/local/sbin/named /usr/sbin/named

    I don't know how good solution this is but at least chkservd shows that bind is up and working.
     
  19. eos1

    eos1 Well-Known Member

    Joined:
    Mar 11, 2003
    Messages:
    175
    Likes Received:
    0
    Trophy Points:
    16
    That's the good one. :)

    Also, you may need to symlink:
    ndc
    rndc
    rndc.conf
    rndc.key
    as well.

    Thanks, Adam(cPanel)!

    Ref:
    Fatal, neither ndc or rndc was found on this server

    Error reloading bind on XXXX: rndc: neither /usr/local/etc/rndc.conf nor /usr/local/etc/rndc.key was found
     
    #19 eos1, Mar 17, 2005
    Last edited: Mar 17, 2005
  20. totaltec

    totaltec Registered

    Joined:
    Jan 10, 2005
    Messages:
    4
    Likes Received:
    0
    Trophy Points:
    1
    anymore news on this i have bind running and thanks to the posts here i didnt restart bind so all is ok.

    any news on the fix or shall i just sim link as posted above
     
Loading...

Share This Page