inodes depleted

[David A. De Graaf]

Inn has used up all the inodes WITHOUT throttling.
It took me a while to discover why I kept getting out-of-space
messages until I tried df -i:

# df /var
Filesystem           1k-blocks      Used Available Use% Mounted on
/dev/sda5              7416060   5339996       76% /var

# df -i /var
Filesystem            Inodes   IUsed   IFree IUse% Mounted on
/dev/sda5             942848  942848       0  100% /var

I realize I must rebuild the filesystem with more inodes,
but the real mystery is why didn't inn-watch throttle?
The default /etc/news/innwatch.ctl contains this line:

!!! ${INNDF} -i . ! lt ! ${INNWATCHSPOOLNODES} ! throttle ! No space (spool inodes)

If I change it to this, throttling does occur:

!!! ${INNDF} -i ${PATHDB} ! lt ! ${INNWATCHSPOOLNODES} ! throttle ! No space (spool inodes)

Apparently, the command    ${INNDF} -i .    is testing the wrong
filesystem, but   ${INNDF} -i ${PATHDB}   is correct.
If this conjecture is correct, the other test, a few lines earlier,
should also be changed to:

!!! ${INNDF} ${PATHDB} ! lt ! ${INNWATCHSPOOLSPACE} ! throttle ! No space (spool)

It seems unnecessarily ambiguous to use . when ${PATHDB} is available.
What subtlety am I missing here?
Shouldn't the default innwatch.ctl be changed?
BTW, I'm using the inn-2.2.1-1 rpm  on a Red Hat 6.1 Linux system.

-- 
	David A. De Graaf    DATIX, Inc.    Hilton Head Is., SC  
	degraaf at rhsnet.com       843-785-3136, -3156 (fax)