/ Register

Join the community today
Register Now

Nginx Nginx unable to start after I Migrate Linode Host Nodes

Discussion in 'Nginx, PHP-FPM & MariaDB MySQL' started by rdan, Aug 30, 2017.

Tags:
Previous Thread Next Thread
Loading...
  1. Aug 30, 2017 #1
    rdan

    rdan Premium Member Premium Member

    4,419
    1,059
    113
    May 25, 2014
    Ratings:
    +1,543
    Local Time:
    1:50 AM
    Mainline
    10.2
    Please fill in any relevant information that applies to you:
    • CentOS Version:CentOS 7 64bit
    • Centmin Mod Version Installed:123.09beta01
    • Nginx Version Installed: 1.13.4
    Code:
    # systemctl status nginx.service
● nginx.service - SYSV: Nginx is an HTTP(S) server, HTTP(S) reverse proxy and IMAP/POP3 proxy server
   Loaded: loaded (/etc/rc.d/init.d/nginx; bad; vendor preset: disabled)
   Active: failed (Result: exit-code) since Wed 2017-08-30 07:57:52 UTC; 1min 1s ago
     Docs: man:systemd-sysv-generator(8)
  Process: 30662 ExecStart=/etc/rc.d/init.d/nginx start (code=exited, status=1/FAILURE)

Aug 30 07:57:52 hostname systemd[1]: Starting SYSV: Nginx is an HTTP(S) server, HTTP(S) reverse proxy and IMAP/POP3 proxy server...
Aug 30 07:57:52 hostname nginx[30662]: Starting nginx: /bin/bash: line 1: 30668 Illegal instruction     /usr/local/sbin/nginx -c /usr/local/nginx/conf/nginx.conf
Aug 30 07:57:52 hostname nginx[30662]: [FAILED]
Aug 30 07:57:52 hostname systemd[1]: nginx.service: control process exited, code=exited status=1
Aug 30 07:57:52 hostname systemd[1]: Failed to start SYSV: Nginx is an HTTP(S) server, HTTP(S) reverse proxy and IMAP/POP3 proxy server.
Aug 30 07:57:52 hostname systemd[1]: Unit nginx.service entered failed state.
Aug 30 07:57:52 hostname systemd[1]: nginx.service failed.

     
  2. Aug 31, 2017 #2
    Matt

    Matt Moderator Staff Member

    795
    354
    63
    May 25, 2014
    Rotherham, UK
    Ratings:
    +537
    Local Time:
    5:50 PM
    1.5.15
    MariaDB 10.2
    Does the nginx config still pass the test? nginx -t ?
     
  3. Aug 31, 2017 #3
    rdan

    rdan Premium Member Premium Member

    4,419
    1,059
    113
    May 25, 2014
    Ratings:
    +1,543
    Local Time:
    1:50 AM
    Mainline
    10.2
    I have fixed it by recompiling Nginx.
    Without custom_config.inc.

    Thanks Matt.
     
  4. Aug 31, 2017 #4
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    What cpu does new server have? Did you do fresh 123.09beta01 install? Or copy over nginx binary from old server?

    Try setting CLANG='y' in persistent config file /etc/centminmod/custom_config.inc and recompile nginx via centmin.sh menu option 4
     
    Last edited: Aug 31, 2017
  5. Aug 31, 2017 #5
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    What was its contents?
     
  6. Aug 31, 2017 #6
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    Oh did you do a Linode clone or migrate via Linode's feature to move entire server data from one host node to another host node ? That could explain the error as nginx and php can be optionally setup to use intel optimised compiler and that is tied to the specific cpu model used on server. So when you use Linode's own migrate feature, you copied everything from one host node to another host node but they had different cpu models, so the nginx and maybe php binary are no longer working for the changed cpu model until you do centmin.sh menu option 4 and 5 recompiles.

    Overlooked that this can be a usage scenario encountered on cloud hosting providers. Will have to add to 123.09beta01 a way to track initial installed cpu model and check if cpu model changes to alert users to recompile nginx and php if required.
     
    • Winner Winner x 1
  7. Aug 31, 2017 #7
    rdan

    rdan Premium Member Premium Member

    4,419
    1,059
    113
    May 25, 2014
    Ratings:
    +1,543
    Local Time:
    1:50 AM
    Mainline
    10.2
    Still the old one I think.
    Both no.

    99% the same with this: gist:dc815286195eece5aff3cdbecd5d9585 · GitHub
     
  8. Aug 31, 2017 #8
    rdan

    rdan Premium Member Premium Member

    4,419
    1,059
    113
    May 25, 2014
    Ratings:
    +1,543
    Local Time:
    1:50 AM
    Mainline
    10.2
    It's because of this I have to do it:
     
  9. Aug 31, 2017 #9
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    Most likely reason is due to different cpu models you could of had E5-2680v3 or E5-2697v4 and switched to E5-2680v2 ? Could also be due to using DEVTOOLSETSIX='y' in customconfig.inc and new server not having GCC 6.2.1 installed yet ?

    what's ouput on new server for these 2 commands
    Code (Text):
    /opt/rh/devtoolset-6/root/usr/bin/gcc --version

    Code (Text):
    /opt/rh/devtoolset-6/root/usr/bin/g++ --version
     
  10. Aug 31, 2017 #10
    rdan

    rdan Premium Member Premium Member

    4,419
    1,059
    113
    May 25, 2014
    Ratings:
    +1,543
    Local Time:
    1:50 AM
    Mainline
    10.2
    It's the same for PHP after migration, it doesn't start.
    But after I recompile it still using the old custom_config it works.
    Nginx only resolve when I recompile it without custom_config first.
     
  11. Aug 31, 2017 #11
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    it's probably one of the custom_config.inc set settings then for yours gist:dc815286195eece5aff3cdbecd5d9585 · GitHub

    for nginx just set these 3 in persistent config /etc/centminmod/custom_config.inc and recompile nginx via centmin.sh menu option 4 and see
    Code (Text):
    CLANG='n'
DEVTOOLSETSIX='y'
NGINX_DEVTOOLSETGCC='y'

    and see if that works, then would have to systematically troubleshoot some of the other custom config settings you set on subsequent nginx recompile tests
     
  12. Aug 31, 2017 #12
    rdan

    rdan Premium Member Premium Member

    4,419
    1,059
    113
    May 25, 2014
    Ratings:
    +1,543
    Local Time:
    1:50 AM
    Mainline
    10.2
    But it works now with those custom config, no issues now.
     
  13. Aug 31, 2017 #13
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    could be devtoolset-6 GCC 6.2.1 wasn't installed immediately after move but was after recompile with no custom_config.inc settings ? check when devtoolset-6 was installed
    Code (Text):
    yum history info devtoolset-6-gcc
     
  14. Aug 31, 2017 #14
    rdan

    rdan Premium Member Premium Member

    4,419
    1,059
    113
    May 25, 2014
    Ratings:
    +1,543
    Local Time:
    1:50 AM
    Mainline
    10.2
     
  15. Aug 31, 2017 #15
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    looks installed so ok there. Strange then
     
  16. Sep 9, 2017 #16
    eva2000

    eva2000 Administrator Staff Member

    37,237
    8,134
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +12,524
    Local Time:
    3:50 AM
    Nginx 1.15.x
    MariaDB 5.5/10.x
    • Like Like x 1
Previous Thread Next Thread
Loading...
..

.