rcj@burl.UUCP (Curtis Jackson) (08/26/85)
Someone posted an article saying that they had suffered articles (files in their news directory [/usr/spool/news, whatever]) that were zero-length files. I, too, have just experienced that problem for the first time; two articles in net.wanted.sources were zero length. My clue is: I have been running 2.10.2 for a long time now, and have never noticed this problem before. Then again, my /usr/spool/news filesystem never ran out of free disk space before, either -- until last night. I believe that somewhere in the vastness of the 2.10.2 software the zero-length articles are being created because the sending system cannot open a temp file. I haven't had time to poke through the source to verify this, but if I find the time to do so before one of you does, I'll post anything I find. -- The MAD Programmer -- 919-228-3313 (Cornet 291) alias: Curtis Jackson ...![ ihnp4 ulysses cbosgd mgnetp ]!burl!rcj ...![ ihnp4 cbosgd akgua masscomp ]!clyde!rcj
ladm@kitc.UUCP (John Burgess) (08/27/85)
In article <829@burl.UUCP> rcj@burl.UUCP (Curtis Jackson) writes: >Someone posted an article saying that they had suffered articles (files in >their news directory [/usr/spool/news, whatever]) that were zero-length >files. I, too, have just experienced that problem for the first time; ... >never ran out of free disk space before, either -- until last night. ... Yup, Curtis, that seems to be the problem all right. We've been losing lots of junk (and probably even a gem or two) because we keep EVERYthing in /usr/spool (which is its OWN filesystem!) Because our feed comes in batched, we can run out of space overnight even when there's over 2000 free blocks at 5PM! I'm triply hit with this storage technique: 1) /usr/spool/uucp/<the batched input> 2) /usr/spool/news/<the received articles> 3) /usr/spool/newslib/log, history (lists articles even when 0 length!) (of course, this doesn't count for much, except when running expire history replicates itself!) Looks like its time for us to A) cut down on how long we retain groups and/or B) enlarge /usr/spool -- John Burgess - Local Tools Administrator ATT-IS Labs, So. Plainfield NJ (HP 1C-221) {most Action Central sites}!kitc!ladm (201) 561-7100 x2481 (8-259-2481)