Why "commBind: Cannot bind socket FD 6 to *:3128:..."?

I run Squid (v2.3S2, on a Linux box) as a transparent proxy. I also
use redirectors, and update them weekly. After the update, I run
"squid -k reconfigure" to reload the redirectors.

This has been running great. Last week, though, I received a message
in cache.log advising me to bump up the number of redirector children.
I edited squid.conf, increasing the count from 5 to 6. I didn't
reconfigure Squid at that time, thinking it would be picked up on the
next weekly redirector reload. Well, that reconfig crashed Squid (see
below).

When I manually restarted Squid the next morning, it had no complaints
at all about having 6 redirector children.

Can anyone suggest a reason that Squid crashed on me?

Thank you.

----- Start cache.log excerpt -----

Restarting Squid Cache (version 2.3.STABLE2)...
FD 6 Closing HTTP connection
FD 13 Closing ICP connection
Cache dir '/var/spool/squid' size remains unchanged at 262144 KB
DNS Socket created on FD 4
Adding nameserver 127.0.0.1 from /etc/resolv.conf
Adding nameserver 24.4.162.33 from /etc/resolv.conf
Adding nameserver 24.4.162.34 from /etc/resolv.conf
helperOpenServers: Starting 6 'squid_redirect' processes
commBind: Cannot bind socket FD 6 to *:3128: (98) Address already in use
storeDirWriteCleanLogs: Starting...
  Finished. Wrote 28562 entries.
  Took 0.4 seconds (75197.0 entries/sec).
FATAL: Cannot open HTTP Port
Squid Cache (Version 2.3.STABLE2): Terminated abnormally.

----- End cache.log excerpt -----

*** Steve Snyder ***
Received on Sun Apr 16 2000 - 11:17:35 MDT