jane@tolerant.UUCP (Jane Medefesser) (10/22/88)
I rebuilt my history file Wednesday night because it was simply huge.
It resulted in the file shrinking about 80%. Fine. NOW, I am getting
duplicate articles. (?) Seriously, my news software isn't rejecting
articles with the same message ID as dups anymore and my spool directory
is filling up FAST!! What have I done? What can I do?
(oh yea - I'm running B news 2.11.3)
============================================
Jane Medefesser uucp: {pyramid,oliveb,ucbvax}!tolerant!jane
Usenet Administrator
Tolerant Systems
San Jose, Ca 95134cole@unix.SRI.COM (Susan Cole) (10/29/88)
In article <2839@tolerant.UUCP> jane@tolerant.UUCP (Jane Medefesser) writes: >I rebuilt my history file Wednesday night because it was simply huge. >It resulted in the file shrinking about 80%. Fine. NOW, I am getting >duplicate articles. (?) Seriously, my news software isn't rejecting >articles with the same message ID as dups anymore and my spool directory >is filling up FAST!! What have I done? What can I do? I dealt with a similar problem by using the -p argument to expire for awhile after the rebuild: expire -e 10 -E 28 -p The manual says that -p "causes expire to use the posting date rather than the received date to expire articles". I assume that in the absence of a "Posted:" header, the "Date:" header is used. Using -p seems to make expire run slower, I assume because expire now has to look inside every article to determine whether to expire it. And you might expire some things early if they wandered around the net a bit before getting to you. This method doesn't stop old articles from being accepted, but they are expired next time expire is run, so they don't annoy users or use disk space for very long. You could also shorten the HISTEXP variable in defs.h and recompile news, but this would make the history file forget about things in whatever short time you'd specified as well as making rnews reject any articles older than that, so whenever you tried to increase HISTEXP again you'd have a similar problem. Also, you don't want HISTEXP to be shorter than DFLTEXP. -- - Susan Cole USENET: ...!hplabs!sri-unix!cole ARPA: cole@unix.sri.com