Welcome to Centmin Mod Community
Become a Member

CentOS 7.x Centminmod beta doesn't connect to network after reboot

Discussion in 'Beta release code' started by jeffwidman, Feb 24, 2015.

  1. jeffwidman

    jeffwidman Active Member

    152
    27
    28
    Dec 3, 2014
    Ratings:
    +51
    Local Time:
    4:49 PM
    Now that George pushed the fix that disables Apache, I was going to migrate a site this weekend to a new RamNode 2gb OpenVZ VPS with Centminmod.


    I started with their CentOS 7 minimal image, had everything installed, did a test migration and everything seemed fine and I could access the Centminmod default page from a web browser. Then for some reason I tried `reboot`. When I tried to SSH back in, I kept getting "no route to host" errors, and the browser page was inaccessible. I emailed RamNode support, and they said:
    I'm afraid that no matter what I tried, I can't get the networking permanently back online. There seems to be some configuration issue with the way CentOS 7 does networking and while I can bring it online by hand, after rebooting the server it doesn't come up anymore.

    I had a hunch it's something in Centminmod that interferes with systemd restarting the networking, so to test I wiped the box and reinstalled a fresh copy of RamNode's CentOS 7 minimal image. Then I ran following commands from bash after SSH'ing into server--these are the only commands I ran from a completely fresh VPS:
    1. yum install vim git which
    2. git clone -b 123.08centos7beta01 --single-branch centminmod/centminmod · GitHub
    3. chmod +x centminmod/centmin.sh
    4. vim centminmod/centmin.sh
      1. set email address
      2. set compile nginx with ipv6 to 'y'
      3. set vim instead of nano
    5. centminmod/centmin.sh
      1. menu screen comes up
      2. exit menu (#22) without installing anything.
    6. try `reboot` - server comes back up just fine.
    7. cd centminmod
    8. ./centmin.sh
      1. at menu, enter #1 to install default centminmod
      2. once that completes, use #22 to exit
    9. Try rebooting server. Fails to come up.
    Rebooted fine before centminmod install, but afterwards SSH attempts always cause "no route to host" errors. I'm not doing anything fancy when rebooting, just typing `reboot` at the command prompt.

    RamNode's VPS control panel includes a serial console for emergency SSH'ing in, and that works, but otherwise the box is inaccessible.

    @eva2000 I just emailed you login creds to the control panel for this box if you want to SSH in using this serial console and review the logs. Also, feel free to reset the box to a fresh image and test doing this yourself using the "reinstall" button from RamNode's control panel.
     
  2. eva2000

    eva2000 Administrator Staff Member

    54,527
    12,211
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +18,780
    Local Time:
    10:49 AM
    Nginx 1.27.x
    MariaDB 10.x/11.4+
    try NOT doing steps 5 and 6 in the above process and see - when centmin.sh first loads it doesn't some background prep expecting menu option 1 to be the first step you run usually
     
  3. jeffwidman

    jeffwidman Active Member

    152
    27
    28
    Dec 3, 2014
    Ratings:
    +51
    Local Time:
    4:49 PM
    Thanks for the suggestion George, happy to try it, but it does seem a bit silly. If I exit without doing anything, nothing should be changed. And if the VPS reboots fine at step 6, then I'd be shocked if it wasn't step 8.1 that was causing the problem.
    I'm also fairly sure the first time I had this issue (before I was trying to isolate the root cause) that I'd skipped step 5 & 6.

    Will try it though and report back with results later today...
     
  4. jeffwidman

    jeffwidman Active Member

    152
    27
    28
    Dec 3, 2014
    Ratings:
    +51
    Local Time:
    4:49 PM
    Just tried the above without steps 5 and 6.

    Issue still exists--after reboot I cannot access via SSH--get a "no route to host" error.

    As before, I am able to connect via the serial console in the RamNode VPS control panel. @eva2000 I left it in that state if you want to login and check the logs you are welcome to (using the creds I emailed you)
     
  5. eva2000

    eva2000 Administrator Staff Member

    54,527
    12,211
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +18,780
    Local Time:
    10:49 AM
    Nginx 1.27.x
    MariaDB 10.x/11.4+
    unfortunately busy right now, but can take a look at it later :)

    if you try restarting CSF Firewall do you get errors and/or can you SSH in ?

    Code:
    service csf restart
    service lfd restart
     
  6. jeffwidman

    jeffwidman Active Member

    152
    27
    28
    Dec 3, 2014
    Ratings:
    +51
    Local Time:
    4:49 PM
    Good idea. Just went in through the serial console:

    Code:
    [root@host1 /]# service csf restart
    Redirecting to /bin/systemctl restart  csf.service
    Job for csf.service canceled.
    [root@host1 /]# service lfd restart
    Redirecting to /bin/systemctl restart  lfd.service
    Job for lfd.service canceled.
    [root@host1 /]#
    
    Just tried SSH'ing in, and it still isn't working unfortunately. I'm not sure it should be saying "job for ___.service canceled" though... is that indicative of a problem?
     
  7. eva2000

    eva2000 Administrator Staff Member

    54,527
    12,211
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +18,780
    Local Time:
    10:49 AM
    Nginx 1.27.x
    MariaDB 10.x/11.4+
    yeah looks like CSF Firewall problem !

    Will look into when I have some free time :)
     
  8. JarylW

    JarylW Active Member

    216
    41
    28
    Jun 19, 2014
    Singapore
    Ratings:
    +103
    Local Time:
    8:49 AM
    Was going to create a new thread, but quite possibly it's the same issue as this.

    @jeffwidman - did you managed to investigate the cause?

    P.S. sorry for necro - thread was on page 1 and didn't think there was a need for 2 separate threads.
     
  9. eva2000

    eva2000 Administrator Staff Member

    54,527
    12,211
    113
    May 24, 2014
    Brisbane, Australia
    Ratings:
    +18,780
    Local Time:
    10:49 AM
    Nginx 1.27.x
    MariaDB 10.x/11.4+
    believe @jeffwidman issue was solved back in March 3 with this commit temp work around for CentOS 7 + OpenVZ and dealing with /tmp mount · centminmod/centminmod@6ae37ad · GitHub

    @JarylW you are using latest .08 beta 02 code right ? if you are probably best to start a new thread as it could be a totally different issue