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.

Clustering and Server Load

Discussion in 'Bind / DNS / Nameserver Issues' started by Tom Pyles, Mar 20, 2004.

  1. Tom Pyles

    Tom Pyles Well-Known Member

    Joined:
    Apr 26, 2002
    Messages:
    254
    Likes Received:
    0
    Trophy Points:
    16
    Question for you folks that are using CPanel's clustering between servers. I recently started using it as it seems many of the bugs have been ironed out. Since setting it up, it seems the load on the master server goes up pretty high when the servers are talking to each other. Under 'top', I see 'dnsadmin' being the culprit.

    I don't know if this is happening because we are just setting these up. I'm hoping things will settle down after everyone is in sync. Is anyone else seeing this happen?

    Our master server normally usually has very little load, even during back-ups and log processing. Just syncing the servers:


    01:00:20 up 3 days, 20:32, 1 user, load average: 3.87, 4.30, 4.26
    98 processes: 92 sleeping, 5 running, 0 zombie, 1 stopped
    CPU states: cpu user nice system irq softirq iowait idle
    total 81.4% 0.0% 18.3% 0.0% 0.1% 0.0% 0.0%
    Mem: 1022476k av, 982604k used, 39872k free, 0k shrd, 139220k buff
    604840k actv, 298056k in_d, 7572k in_c
    Swap: 1052248k av, 102648k used, 949600k free 568232k cached

    PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME CPU COMMAND
    18335 root 25 0 3772 3772 1464 R 49.5 0.3 0:03 0 dnsadmin
    18356 root 25 0 3780 3780 1468 R 32.9 0.3 0:01 0 dnsadmin
    18352 root 23 0 11320 11M 5416 S 1.7 1.1 0:00 0 whostmgr2
    18360 root 24 0 5340 5340 2472 S 0.9 0.5 0:00 0 dnsadmin
    31905 cpanel 15 0 2776 2668 1548 S 0.3 0.2 0:27 0 stunnel-4.04loc
    13812 named 25 0 40104 38M 1952 S 0.1 3.8 30:25 0 named
    21569 root 15 0 1276 1276 900 R 0.1 0.1 0:09 0 top
    18351 root 20 0 5460 5344 2820 S 0.1 0.5 0:00 0 cpsrvd
    1 root 15 0 488 460 432 S 0.0 0.0 0:04 0 init
    2 root 15 0 0 0 0 SW 0.0 0.0 0:01 0 keventd
    3 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kapmd
    4 root 35 19 0 0 0 SWN 0.0 0.0 0:00 0 ksoftirqd/0
    7 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 bdflush
    5 root 15 0 0 0 0 SW 0.0 0.0 0:04 0 kswapd
    6 root 15 0 0 0 0 SW 0.0 0.0 0:01 0 kscand
    8 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kupdated
    9 root 25 0 0 0 0 SW 0.0 0.0 0:00 0 mdrecoveryd
    13 root 15 0 0 0 0 SW 0.0 0.0 1:04 0 kjournald
    69 root 25 0 0 0 0 SW 0.0 0.0 0:00 0 khubd
    583 root 15 0 0 0 0 SW 0.0 0.0 0:00 0 kjournald
    3378 root 15 0 308 288 224 R 0.0 0.0 0:34 0 syslogd
    3382 root 15 0 200 188 144 S 0.0 0.0 0:00 0 klogd
    3499 root 15 0 784 388 300 S 0.0 0.0 0:00 0 cupsd
     
  2. Higgins

    Higgins Well-Known Member

    Joined:
    Jan 31, 2003
    Messages:
    82
    Likes Received:
    0
    Trophy Points:
    6
    Yes we have the same Problem, adding a DNS Zone and the Load on every Cluster Server Jumps High
     
  3. phantom

    phantom Well-Known Member

    Joined:
    Sep 17, 2002
    Messages:
    123
    Likes Received:
    0
    Trophy Points:
    16
    Same here. It's crashing my dual xeon box.
     
  4. Higgins

    Higgins Well-Known Member

    Joined:
    Jan 31, 2003
    Messages:
    82
    Likes Received:
    0
    Trophy Points:
    6
    This drives me realy NUTS. Why are you adding things wich crashing our Servers and didnt work propperly ???? No fix for that no answer on the Ticket nothing .... Great Job Darkorb :(
     
  5. web12

    web12 Well-Known Member

    Joined:
    Nov 20, 2002
    Messages:
    240
    Likes Received:
    0
    Trophy Points:
    16
    Is this still the case for other people? I have noticed a considerable increase in load on my master server since turning over to the Clustering. Im just wondering if I have it set up correctly.

    The way I have it is so that all updates to all servers are made through the one master server. Would it lessen the load if updates were done in a chain? So that a change made on the master server spoke to server 2, which then in turn spoke to server 3 and so on...

    Anyone else know if this is the correct way of doing things or not?

    thanks
     
  6. raventec

    raventec Well-Known Member

    Joined:
    Apr 19, 2003
    Messages:
    120
    Likes Received:
    0
    Trophy Points:
    16
    I just noticed mine freaked as well. came looking for answers. :(
     
  7. web12

    web12 Well-Known Member

    Joined:
    Nov 20, 2002
    Messages:
    240
    Likes Received:
    0
    Trophy Points:
    16
    One other question... Ive read a few posts by people saying that the DNS cluster setup should be done using the DNSOnly version of WHM on a DNS Only server, but can anyone recommend a minimum spec for a server running just DNSONLY?
     
  8. xunkNownx

    xunkNownx Registered

    Joined:
    Feb 29, 2004
    Messages:
    1
    Likes Received:
    0
    Trophy Points:
    1
    I had a problem where when I went to change any dns zones or do anything dns related my cpu would peg out at 99% for about 3 min then finally load the zones. I disabled dns clustering, restarted bind and everything is back to normal now. I am using:

    WHM 9.1.0
    Cpanel 9.1.0-C71

    RedHat 9

    Thank you web12 for the comment about using Clusting with the dns versions of Cpanel only.
     
  9. Higgins

    Higgins Well-Known Member

    Joined:
    Jan 31, 2003
    Messages:
    82
    Likes Received:
    0
    Trophy Points:
    6
    And another Thanks to the Developers today the Clustering System has crashed 10 Servers at once.

    I tried to Syncronise the DNS records on all Servers and be advised "NEVER DO THAT" Load on all Serevr gone Rocket High (150 and above) and this Shit never ends it seems that this has run as a Loop Process on server had finished 10 Minutes Later he did the same from the beginning. All Named Zones from now on are 5times in the named.conf i had to reboot all Servers and Restore a Backup off all named.conf


    I can tell you this was real Fun !!!!!
     
  10. phantom2

    phantom2 Well-Known Member

    Joined:
    Jan 3, 2004
    Messages:
    59
    Likes Received:
    0
    Trophy Points:
    6
    I think I might have a solution but I'm not sure.

    I just got 2 new dell servers just for DNS clustering but I have many unanswered questions.

    The old way was the trustkey relationship which turns the machine into a slave machine. It could be that we have to unslave it /scripts/unslave
    but I'm not sure. I think that's why clustering is causing the crashing. Do we need to delete the old trust key relationships by unslaving the mahines?


    I have a real mess on my hands. I was using a master name server (ns1) and all of my other machines connected to it using the trustkey relationship. For redundancy, I activated a name server on each slave machine. Example: ns5, ns6, ns7, etc...

    Now I have my 2 new Dell name servers for NS1 and NS2.
    I'm stuck until I can figure out if I'm supposed to unslave the machines. It does not mention anything about it in the DOCS but it seems logical to unslave my machines since they will not be slaves anymore, they will part of a cluster right?

    2. Some of my clients are using ns1 and ns5 or ns1 and ns6. Now everybody is going to be using ns1 and ns2. How are all of the zone files going to get updates. I dont want to start this cluster and then realize that I have to manually edit a ton of DNS zones.

    Aside from my problem, I do believe the crashing is caused by the machines still using the trustkey relationships. Can anyone confirm this?

    Also, if anyone knows the answers to 1 and 2 please let me know what to do.

    Thanks!
     
  11. euselect

    euselect Well-Known Member

    Joined:
    Aug 3, 2003
    Messages:
    51
    Likes Received:
    0
    Trophy Points:
    6
    Much the same load issues here with 549 zones loaded over 3 servers , ticket has been opened , if anyone found a solution i am most interested

    For now i have echoed > dnsadmin binary , to stop it loading up my servers.
     
  12. wheimeng

    wheimeng Active Member

    Joined:
    Mar 4, 2003
    Messages:
    32
    Likes Received:
    0
    Trophy Points:
    6
    yea, anyone found any solution?
     
  13. mramos

    mramos Member

    Joined:
    Dec 10, 2002
    Messages:
    23
    Likes Received:
    0
    Trophy Points:
    1
    Same Problem

    Hi !

    I'm have the same problem with load average.

    Yesterday was at 0.1 and after the cluster configuration are on 7.5

    Any sugestion ?

    Tks !
     
  14. DWHS.net

    DWHS.net Well-Known Member
    PartnerNOC

    Joined:
    Jul 28, 2002
    Messages:
    1,569
    Likes Received:
    6
    Trophy Points:
    38
    Location:
    LA, Costa RIca
    cPanel Access Level:
    Root Administrator
    This sounds tricky, like the redundancy is looping somehow. I think you will have to wait for the entire script to get updated by cpanel or maybe try edge to see if it helps as a last resort.
     
  15. andyf

    andyf Well-Known Member

    Joined:
    Jan 7, 2002
    Messages:
    246
    Likes Received:
    0
    Trophy Points:
    16
    Location:
    UK
    I think euselect said this was fixed in the latest edge
     
  16. cyon

    cyon Well-Known Member
    PartnerNOC

    Joined:
    Jan 15, 2003
    Messages:
    320
    Likes Received:
    0
    Trophy Points:
    16
    #16 cyon, Aug 26, 2004
    Last edited: Aug 26, 2004
  17. Finley Ave

    Finley Ave Active Member

    Joined:
    Feb 28, 2004
    Messages:
    37
    Likes Received:
    0
    Trophy Points:
    6
    Location:
    San Ramon, CA
    It looks like nobody's had any trouble with this for a couple of years, but I'm having a problem like this. I have a dns only server and a regular whm server, with less than 20 records, and whenever I sync records, my load goes over 10 and the server is unresponsive for several minutes, showing processes whosmgr and dnsadmin using a lot of cpu. Load goes up only on the regular whm server, not on the dns only, and it doesn't matter which side I initiate a synchronization from. I'm running latest release (r113). I've looked at everything mentioned in this thread. Any ideas?
     
  18. DWHS.net

    DWHS.net Well-Known Member
    PartnerNOC

    Joined:
    Jul 28, 2002
    Messages:
    1,569
    Likes Received:
    6
    Trophy Points:
    38
    Location:
    LA, Costa RIca
    cPanel Access Level:
    Root Administrator
    I know it's not much help, but we just do it manually. Last time I did the sync option is threw off a couple hundred dns zones that we had to manually correct.
     
  19. Finley Ave

    Finley Ave Active Member

    Joined:
    Feb 28, 2004
    Messages:
    37
    Likes Received:
    0
    Trophy Points:
    6
    Location:
    San Ramon, CA
    I can live with that. It kind of defeats the purpose of dnsonly, but whatever works. If I want to do that, I need the zone file from /var/named/ and also the entry in named.conf; is that all?
     
  20. DWHS.net

    DWHS.net Well-Known Member
    PartnerNOC

    Joined:
    Jul 28, 2002
    Messages:
    1,569
    Likes Received:
    6
    Trophy Points:
    38
    Location:
    LA, Costa RIca
    cPanel Access Level:
    Root Administrator
    Yep, that's all.

    I just updated those files on the new dns cluster from the old servers that I brought into the cluster. And all new servers and accounts work fine with automatically connecting.

    It was just syncing older servers that seems to be risky.
     
Loading...

Share This Page