[Bucardo-general] Test Failures: Serialized Isolation

[David E. Wheeler]

On Oct 23, 2012, at 5:22 PM, David E. Wheeler <david at justatheory.com> wrote:

> I think the key point here, though, is that in the first example, only the pgtest5 sync is sent. In the second, both the pgtest5 and the samedb syncs are sent to the same kid! However, the name of the sync is sent to the kid when it starts, so this is the line that triggers the run:
> 
>                elsif ($name eq "run_$syncname") {
>                    $dorun = 1;
>                }
> 
> When both syncs are set, what is the value of $syncname? It looks as though $syncname is set at KID start time. If so, we should either:
> 
> 1. Fix the bug that sends the samedb sync to the wrong KID; or
> 2. Fix the bug where the KID thinks it manages only one sync
> 
> Does that sound right?

I put the $syncname into my Dumper line, and in a failed run saw this:

(68840) [Tue Oct 23 17:30:24.917 2012] #2667 KID For samedb, got: $BUCARDO1 = {
  'run_pgtest5' => {
    'count' => 1,
    'firstpid' => 68863,
    'pid' => {
      '68863' => 1
    }
  },
  'run_samedb' => {
    'count' => 1,
    'firstpid' => 68833,
    'pid' => {
      '68833' => 1
    }
  }
};

In a successful run, I see:

(69298) [Tue Oct 23 17:32:07.934 2012] #2667 KID For pgtest5, got: $BUCARDO1 = {
  'run_pgtest5' => {
    'count' => 1,
    'firstpid' => 69294,
    'pid' => {
      '69294' => 1
    }
  },
  'run_samedb' => {
    'count' => 1,
    'firstpid' => 69259,
    'pid' => {
      '69259' => 1
    }
  }
};

And also:

(69266) [Tue Oct 23 17:32:08.505 2012] #2667 KID For samedb, got: $BUCARDO1 = {
  'run_pgtest5' => {
    'count' => 1,
    'firstpid' => 69294,
    'pid' => {
      '69294' => 1
    }
  },
  'run_samedb' => {
    'count' => 1,
    'firstpid' => 69259,
    'pid' => {
      '69259' => 1
    }
  }
};

Both then appear again later!

So maybe the KID responsible for the pgtest5 sync isn't running?

Thanks,

David