Articles in this section

See more

High amount of sw-engine processes in process list hung on config.lock file

Avatar
Marc López
Updated
Follow

Applicable to:

  • Plesk Onyx for Linux

Note: This article has the reference to the issue with the fix available:

Please consider updating your server:

Symptoms

  • Scheduled backups are not created.

  • Plesk utilities, e.g.:

    # plesk login

    hang.

  • There are lot of sw-engine processes:

    # ps -e fwww | grep sw-engine
    6846 ? S 2:37 _ sw-engine-fpm: pool plesk
    6890 ? S 2:37 _ sw-engine-fpm: pool plesk
    6903 ? S 2:36 _ sw-engine-fpm: pool plesk
    6904 ? S 2:36 _ sw-engine-fpm: pool plesk
    6918 ? S 2:36 _ sw-engine-fpm: pool plesk
    6920 ? S 2:36 _ sw-engine-fpm: pool plesk
    6921 ? S 2:37 _ sw-engine-fpm: pool plesk
    6924 ? S 2:36 _ sw-engine-fpm: pool plesk
    6936 ? S 2:37 _ sw-engine-fpm: pool plesk
    6939 ? S 2:36 _ sw-engine-fpm: pool plesk

    OR

    # ps -e fwww | grep sw-engine

    24695 ? Zs 0:00 | _ [sw-engine] <defunct>
    24875 ? Zs 0:00 | _ [sw-engine] <defunct>
    29217 ? Zs 0:00 | _ [sw-engine] <defunct>
    31327 ? Zs 0:00 | _ [sw-engine] <defunct>
    15254 ? Zs 0:00 | _ [sw-engine] <defunct>
    16243 ? Zs 0:00 | _ [sw-engine] <defunct>
    19463 ? Zs 0:00 | _ [sw-engine] <defunct>
    20472 ? Zs 0:00 | _ [sw-engine] <defunct>
    13502 ? Zs 0:00 | _ [sw-engine] <defunct>
    15067 ? Zs 0:00 | _ [sw-engine] <defunct>

  • The strace output of such processes contains the following:

    # strace -p6846

    <...>

    restart_syscall(<... resuming interrupted futex ...>) = -1 ETIMEDOUT (Connection timed out)
    fcntl(9, F_SETLKW, {l_type=F_RDLCK, l_whence=SEEK_SET, l_start=0, l_len=0}) = 0
    fcntl(7, F_SETLKW, {l_type=F_WRLCK, l_whence=SEEK_SET, l_start=0, l_len=0}) = 0
    fcntl(7, F_SETLK, {l_type=F_UNLCK, l_whence=SEEK_SET, l_start=0, l_len=0}) = 0
    fcntl(7, F_SETLKW, {l_type=F_WRLCK, l_whence=SEEK_SET, l_start=0, l_len=0}) = 0
    fcntl(7, F_SETLK, {l_type=F_UNLCK, l_whence=SEEK_SET, l_start=0, l_len=0}) = 0
    umask(02) = 022
    stat("/run/lock/lmlib/container_locks/plesk.0/node.1/service", {st_mode=S_IFDIR|S_ISGID|0775, st_size=60, ...}) = 0
    umask(022) = 02
    umask(0111) = 022
    creat("/run/lock/lmlib/container_locks/plesk.0/node.1/service/config.lock", 0777) = 10
    close(10) = 0
    umask(022) = 0111

  • In case of Multi-Server installation, Service Nodes are going offline periodically during the Daily Maintenance Task execution.

Cause

This is Plesk bug with ID #PPPM-8929 that is fixed in Plesk Onyx 17.8 MU 32.

Resolution

Upgrade to Plesk Onyx 17.8 or install latest Plesk updates in case already running Plesk Onyx 17.8.

Note: In case it is not possible, see Additional Information section at the bottom.

Workaround for older Plesk versions

  1. Connect to the server via SSH.

  2. Kill all sw-engine processes:

    # pkill sw-engine

  3. Restart sw-cp-server and sw-engine services:

    # service sw-engine restart
    # service sw-cp-server restart

  4. Uninstall New Relic extension:

    # plesk bin extension --uninstall new-relic

Additional Information

In case upgrade is not an option, try using the ApiCli.zip from the attachment.

Note that the fix was not fully tested yet was working fine on some installations.

# wget https://support.plesk.com/hc/en-us/article_attachments/360010352974/ApiCli.zip
# unzip ApiCli.zip
# mv /usr/local/psa/admin/plib/pm/ApiCli.php{,.bkp}
# cp -a ApiCli.php /usr/local/psa/admin/plib/pm/

Comments

2 comments

Sort by Date Votes
  • Avatar
    Csteamleads

    You will not be able to log into Plesk if you kill all sw-engine processes. You'll likely need to start/restart it as well:

    pkill sw-engine; systemctl restart sw-engine

    or

    pkill sw-engine; service sw-engine restart

    0
    Permalink
  • Avatar
    Ivan Postnikov

    Hello @Csteamleads,

    thank you for noticing. The article will be reviewed and updated soon.

    0
    Permalink

Please sign in to leave a comment.

Have more questions? Submit a request

Related articles