trouble restarting a server - Mailing list pgsql-admin

Hi all, I'm having some trouble restarting a long-running server. Basically, pg_ctl can't kill any processes waiting on a notify interrupt. Here's what happened after trying a "stop -m fast"

[root@vero ~]# ps axvw | grep post
 3606 ?        S      3:30     23  3260 41575 6644  0.1 /usr/bin/postgres -D /postgres/db-8.2
 4293 ?        Ss     0:17      0  3260 40551 27004  0.6 postgres: writer process
 4294 ?        Ss     0:00      0  3260  9363 1744  0.0 postgres: stats collector process
14508 ?        Ss     0:01      0  3260 41195 29356  0.7 postgres: me csdb zim(48570) notify interrupt
20701 ?        Ss     0:01      0  3260 41127 29384  0.7 postgres: others csdb chef(45882) notify interrupt
31179 ?        Ss     0:01      0  3260 41127 29428  0.7 postgres: others csdb yfandes(33113) notify interrupt
 7342 ?        Ss     0:03      0  3260 41723 30584  0.7 postgres: others csdb reed(54823) notify interrupt
 7343 ?        Ss     0:02      0  3260 41127 29396  0.7 postgres: others csdb reed(54824) notify interrupt
 7404 ?        Ss     0:02      0  3260 41127 29392  0.7 postgres: others csdb reed(54830) notify interrupt
29537 ?        Ss     0:02      0  3260 41127 29404  0.7 postgres: others csdb stupid(60028) notify interrupt
26671 pts/0    S+     0:00      1    19  7976 1700  0.0 su postgres -c /s/postgresql-8.2/bin/pg_ctl stop -m fast -D /scratch.1/postgres/csdb-8.2 -l /dev/null
26672 pts/0    S+     0:00      1   283  5556 1180  0.0 -csh -c /usr/bin/pg_ctl stop -m fast -D /scratch.1/postgres/csdb- 8.2 -l /dev/null
26673 pts/0    S+     0:00      0    23  6916 1440  0.0 /usr/bin/pg_ctl stop -m fast -D /postgres/db-8.2 -l /dev/null
26713 ttyS0    S+     0:00      0    71  5016  644  0.0 grep post

Any idea how postgres gets into this state, or how it can be resolved short of a system reboot or a "stop -m immediate"? Thanks much.

Peter