Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-9961

Garbd segfaults when using a configuration file on RH6

    XMLWordPrintable

Details

    Description

      On a Redhat 6.7 installation we see Garbd segfaulting when we use a configuration file when starting Garbd, on Redhat 7.1 we don't see this issue happening.

      [user@n1 ~]$ sudo /usr/bin/garbd --cfg /etc/garbd.cnf
      [user@n1 ~]$ echo $?
      139
      

      Nothing gets logged to the garbd log file.
      If we provide all the options inside the configuration file manually on the command line like this everything is fine:

      sudo /usr/bin/garbd  -g my_wsrep_cluster -a "gcomm://10.10.16.12:4567,10.10.16.13:4567" -o "gmcast.listen_addr=tcp://0.0.0.0:4567" -l /var/log/garbd.log
      

      The configuration files get generated by ClusterControl for both OS-es and contain the exact same information:

      [sudo@n1 ~]$ sudo cat /etc/garbd.cnf
      address = gcomm://10.10.16.12:4567,10.10.16.13:4567
      group = my_wsrep_cluster
      options = gmcast.listen_addr=tcp://0.0.0.0:4567
      log = /var/log/garbd.log
      

      Also the garbd.log file exists and has the right permissions: if we start up Garbd with all the commandline parameters it is fine.

      Using strace gave us the folowing last few lines:

      open("/etc/garbd.cnf", O_RDONLY)        = 3
      read(3, "address=gcomm://10.10.16.12:4567"..., 8191) = 143
      read(3, "", 8191)                       = 0
      close(3)                                = 0
      open("/var/log/garbd.log", O_WRONLY|O_CREAT|O_APPEND, 0666) = 3
      fstat(3, {st_mode=S_IFREG|0644, st_size=11312, ...}) = 0
      mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5bcc8b9000
      fstat(3, {st_mode=S_IFREG|0644, st_size=11312, ...}) = 0
      lseek(3, 11312, SEEK_SET)               = 11312
      gettimeofday({1461197582, 39668}, NULL) = 0
      open("/etc/localtime", O_RDONLY)        = 4
      fstat(4, {st_mode=S_IFREG|0644, st_size=118, ...}) = 0
      fstat(4, {st_mode=S_IFREG|0644, st_size=118, ...}) = 0
      mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5bcc8b8000
      read(4, "TZif2\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\1\0\0\0\1\0\0\0\0"..., 4096) = 118
      lseek(4, -62, SEEK_CUR)                 = 56
      read(4, "TZif2\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\1\0\0\0\1\0\0\0\0"..., 4096) = 62
      close(4)                                = 0
      munmap(0x7f5bcc8b8000, 4096)            = 0
      write(3, "2016-04-21 00:13:02.039  INFO: C"..., 69) = 69
      --- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0xfffffffffffffff8} ---
      +++ killed by SIGSEGV +++
      

      I have added the full strace to the report.

      The package we install:
      galera-25.3.15-1.rhel6.el6.x86_64

      Attachments

        Issue Links

          Activity

            People

              nirbhay_c Nirbhay Choubey (Inactive)
              art.severalnines Art van Scheppingen
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved:

                Git Integration

                  Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.