hot standby waiting for invalid wal file

Dear List

Since 9.0 comes out, I had been successfully using wal file shipping
readonly hotstandby feature in pg. I am currently trying to
reestablish a readonly slave after master node failure.

The newly established master server is working properly. I have
verified that WAL file shipping is properly working.

Next I do

psql -c "select pg_start_backup('label')"

then I rsync $PGDATA to slave

Exact command is (nice rsync -avz --delete -F --exclude
postgresql.conf --exclude archive_status --exclude
backup_label --exclude postmaster.pid -e ssh data/
pgslave:./data/)

followed by

psql -c "select pg_stop_backup()"

Got "NOTICE: pg_stop_backup complete, all required WAL segments have
been archived"

I then make sure that standby $PGDATA contains relevant
postgresql.conf and recovery.conf

Then start pg on slave.

But slave server never finishes recovery because it is waiting for old
WAL file 000000010000012C00000097 By old I mean the file name seems
to be earlier than what is newly generated WAL files

I have repeated above procedure a few times. Everytime slave keeps
wating for the same 000000010000012C00000097

Can someone give any suggestions on how to solve this?

Sincerely

Wu