CTDB Assigned IP not on an interface

Max DiOrio Max.DiOrio at ieeeglobalspec.com
Thu Sep 26 23:55:07 UTC 2019


Thank you Martin!  I spent hours trying to get this working today.  My google-fu was weak today.  I didn't find anything about the format change or the disabling of the legacy 10.interface

BTW, that strace was of the ctdb master process.

Sorry about posting here.  I googled ctdb mailing list and this is the only one I came up with.  Again, google-fu failing me today.

You rock!

Max


On 9/26/19, 7:16 PM, "Martin Schwenke" <martin at meltin.net> wrote:

    NOTE: This email originated from outside of the organization.
    
    
    Hi Max,
    
    On Thu, 26 Sep 2019 19:04:31 +0000, Max DiOrio via samba-technical
    <samba-technical at lists.samba.org> wrote:
    
    > Additional info - when running an strace on the process:
    >
    > epoll_wait(11, [{EPOLLIN, {u32=2179214400, u64=94542999333952}}], 1, 114) = 1
    > ioctl(24, FIONREAD, [96])               = 0
    > read(24, "`\0\0\0BDTC\1\0\0\0\352k\35w\7\0\0\0\0\0\0\0\2\0\0\0W\24\1\0"..., 96) = 96
    > stat("/etc/localtime", {st_mode=S_IFREG|0644, st_size=3519, ...}) = 0
    > write(5, "2019/09/26 14:59:36.055468 ctdbd"..., 130) = 130
    > socket(AF_INET, SOCK_STREAM, IPPROTO_TCP) = 25
    > bind(25, {sa_family=AF_INET, sin_port=htons(0), sin_addr=inet_addr("10.85.136.108")}, 16) = -1 EADDRNOTAVAIL (Cannot assign requested address)
    > close(25)                               = 0
    > stat("/etc/localtime", {st_mode=S_IFREG|0644, st_size=3519, ...}) = 0
    > write(5, "2019/09/26 14:59:36.055731 ctdbd"..., 118) = 118
    > stat("/etc/localtime", {st_mode=S_IFREG|0644, st_size=3519, ...}) = 0
    > write(5, "2019/09/26 14:59:36.055848 ctdbd"..., 153) = 153
    > write(21, "0\0\0\0BDTC\1\0\0\0\352k\35w\10\0\0\0\2\0\0\0\0\0\0\0W\24\1\0"..., 48) = 48
    
    Not sure what process you're running strace on here.  A higher
    value for -s might have give enough context in the log messages
    being written in that snippet.
    
    However, I think I can guess the answer... see below...
    
    > On 9/26/19, 1:00 PM, "samba-technical on behalf of Max DiOrio via samba-technical" <samba-technical-bounces at lists.samba.org on behalf of samba-technical at lists.samba.org> wrote:
    
    >     CTDB has been working great for us the last few weeks.  Today, I
    > did a yum update – and now things aren’t working.  The public IP
    > assigned is not actually being added to the interface, but CTDB
    > claims the nodes are healthy.
    
    > [...]
    
    >     Currently running ctdb.x86_64    4.9.1-6.el7. on Centos 7.7.1908
    
    I'm guessing that this was an upgrade to 4.9 from a previous version.
    In 4.9 CTDB's configuration changed a lot and so did the event script
    handling.
    
    Try running:
    
    # ctdb event script list legacy
    
    See if 10.interface is listed and has a '*' next to it to indicate
    whether it is enabled.  If it is listed but it is not enabled then:
    
    # ctdb event script enable legacy 10.interface
    
    will enable it.
    
    You may want to check that all the event scripts you use are enabled as
    expected.
    
    There is an example configuration migration script
    (config_migrate.sh) with CTDB in version 4.9.  It is pretty good but we
    left it as an example because we didn't want to drown in bugs that
    might be present in the script.  I don't know if CentOS 7 ships it.  If
    not, you can get it at:
    
      https://git.samba.org/?p=samba.git;a=blob_plain;f=ctdb/doc/examples/config_migrate.sh;hb=refs/heads/v4-9-stable
    
    If I haven't guessed right then please post your config (ctdb.conf,
    nodes file, public addresses file) and the output of:
    
    # ctdb event script list legacy
    
    If none of this works, since you seem to be awake now, you could jump
    on the #ctdb IRC channel on Freenode and look for me there.  I'll be in
    and out for the next couple of hours.
    
    Thanks...
    
    peace & happiness,
    martin
    
    p.s. As a user level question, this would have been more appropriate on
         the "samba" list, rather than samba-technical.   :-)
    



More information about the samba-technical mailing list