accounting bug

[Ralf Baechle <ralf@uni-koblenz.de>]

Zmailer's internal accounting as displayed by mailq -Q seems to be a bit
broken.  This happend with the CVS snapshot of November 23 on RH 6.2/i386.

[ralf@lappi ralf]$ mailq -Q
smtp/*/0
    smtp/oss.sgi.com/0		R=2   A=17 P={9811} HA={1}s FA={3}s OF={2} S={STUFF} UF=0 QA=2h19m46s
	Threads:    1 Msgs:     2 Procs:   2 Idle:   1 Plim:  50 Flim: 150 Tlim: 1
Kids: 2  Idle:  1  Msgs:   1  Thrds:   1  Rcpnts:    2  Uptime: 2h35m19s
Msgs in 8 out 5 stored 1 Rcpnts in 7 out 5 stored 2
[ralf@lappi ralf]$

Now I start a ETRN queue run in a separate window and a few seconds later:

[ralf@lappi ralf]$ mailq -Q
smtp/*/0
	Threads:    0 Msgs:     0 Procs:   2 Idle:   2 Plim:  50 Flim: 150 Tlim: 1
Kids: 2  Idle:  2  Msgs:  -1  Thrds:   0  Rcpnts:    0  Uptime: 2h35m30s
Msgs in 8 out 7 stored 4294967295 Rcpnts in 7 out 7 stored 0
[ralf@lappi ralf]$

4294967295 is 0xffffffff or -1.

And a bit later:

[ralf@lappi ralf]$ mailq -Q
Kids: 0  Idle:  0  Msgs:  -2  Thrds:   0  Rcpnts:    0  Uptime: 2h58m34s
Msgs in 9 out 7 stored 4294967294 Rcpnts in 7 out 7 stored 0
[ralf@lappi ralf]$

4294967294 is 0xfffffffe or -2.

Hard to deduce the previous queue content but at that point there are no
more files left in the queue and transport directories.

Btw, is there a way to start a full queue run?  ETRN allows to rune queue
runs only destination per destination so isn't what I want on the dial
machine which I use to evaluate zmailer snapshots before use on real
servers.

  Ralf