From news@news.msfc.nasa.gov Mon Mar 31 15:33:56 EST 1997 Article: 31434 of news.software.nntp Path: news.math.psu.edu!news.ems.psu.edu!news3.cac.psu.edu!howland.erols.net!europa.clark.net!news.msfc.nasa.gov!not-for-mail From: news@news.msfc.nasa.gov (Unknown News Administrator) Newsgroups: news.software.nntp Subject: Re: Duplicate articles with INN (bug?) Date: 31 Mar 1997 19:18:01 GMT Organization: NASA/MSFC Lines: 166 Sender: news@hammer.msfc.nasa.gov Message-ID: <5hp2l9$jqa$1@hammer.msfc.nasa.gov> References: <5heo2g$eol$1@sun579.rz.ruhr-uni-bochum.de> <5hh1f6$64k$1@sun579.rz.ruhr-uni-bochum.de> <5hhpre$60$1@news.gv.tsc.tdk.com> NNTP-Posting-Host: hammer.msfc.nasa.gov Xref: news.math.psu.edu news.software.nntp:31434 gdonl@tsc.tdk.com (Don Lewis) writes: >INN 1.5, FreeBSD 2.1.6, w/o MMAP, with a bunch of feeds. I didn't see >any duplicate article files though. It appears that the duplicate history >entries are added when innd rejects articles with "437 Duplicate article" >responses. This does indeed seem to be the problem if REMEMBER_TRASH is set to DO. Below is a patch that reflects all of the changes I have made to my working copy of art.c since INN-1.5.1 came out. It does three things: 1. Some better checking for bad headers. I didn't come up with this one and can't seem to remember where it came from. INN-workers list, maybe? 2. Check hop count *before* checking for path exclusion. Mainly done as an experiment to see if it would produce a noticeable improvement in performance. It might, if you have newsfeeds entries with H flags and looonnng lists of path components to exclude. Really can't hurt. 3. Don't write a history record for articles rejected because they are duplicates. I would have separated out and cleaned up these patches, but it's been a busy day. *** innd/art.c.orig Tue Dec 17 08:40:40 1996 --- innd/art.c Mon Mar 31 12:02:51 1997 *************** *** 689,696 **** *errorp = "EOF in headers"; return NULL; case ':': ! if (colon == NULL) colon = out; break; } if ((*out++ = *in++) == '\n' && !ISWHITE(*in)) --- 689,701 ---- *errorp = "EOF in headers"; return NULL; case ':': ! if (colon == NULL) { colon = out; + if (start == colon) { + *errorp = "Field without name in header"; + return NULL; + } + } break; } if ((*out++ = *in++) == '\n' && !ISWHITE(*in)) *************** *** 1590,1598 **** --- 1595,1608 ---- /* Too small for the site. */ continue; + /* if ((!sp->IgnorePath && ListHas(hops, sp->Name)) || (sp->Hops && hopcount > sp->Hops) || (sp->Groupcount && Groupcount > sp->Groupcount)) + */ + if ((sp->Hops && hopcount > sp->Hops) + || (!sp->IgnorePath && ListHas(hops, sp->Name)) + || (sp->Groupcount && Groupcount > sp->Groupcount)) /* Site already saw the article; path too long; or too much * cross-posting. */ continue; *************** *** 1738,1743 **** --- 1748,1754 ---- char ControlWord[SMBUF]; int ControlHeader; int oerrno; + int adupe; #if defined(DO_PERL) char *perlrc; #endif /* DO_PERL */ *************** *** 1767,1774 **** Data.TimeReceivedLength = strlen(Data.TimeReceived); /* A duplicate? */ ! if (error == NULL && HIShavearticle(Data.MessageID)) error = "Duplicate article"; /* And now check the path for unwanted sites -- Andy */ for( j = 0 ; ME.Exclusions && ME.Exclusions[j] ; j++ ) { --- 1778,1788 ---- Data.TimeReceivedLength = strlen(Data.TimeReceived); /* A duplicate? */ ! adupe = 0; ! if (error == NULL && HIShavearticle(Data.MessageID)) { error = "Duplicate article"; + adupe = 1; + } /* And now check the path for unwanted sites -- Andy */ for( j = 0 ; ME.Exclusions && ME.Exclusions[j] ; j++ ) { *************** *** 1784,1792 **** (void)sprintf(buff, "%d %s", NNTP_REJECTIT_VAL, error); ARTlog(&Data, ART_REJECT, buff); #if defined(DO_REMEMBER_TRASH) ! if (Mode == OMrunning && !HISwrite(&Data, "")) ! syslog(L_ERROR, "%s cant write history %s %m", ! LogName, Data.MessageID); #endif /* defined(DO_REMEMBER_TRASH) */ ARTreject(buff, article); return buff; --- 1798,1808 ---- (void)sprintf(buff, "%d %s", NNTP_REJECTIT_VAL, error); ARTlog(&Data, ART_REJECT, buff); #if defined(DO_REMEMBER_TRASH) ! if (adupe == 0) { ! if (Data.MessageID && Mode == OMrunning && !HISwrite(&Data, "")) ! syslog(L_ERROR, "%s cant write history %s %m", ! LogName, Data.MessageID); ! } #endif /* defined(DO_REMEMBER_TRASH) */ ARTreject(buff, article); return buff; *************** *** 1799,1805 **** syslog(L_NOTICE, "rejecting[perl] %s %s", HDR(_message_id), buff); ARTlog(&Data, ART_REJECT, buff); #if defined(DO_REMEMBER_TRASH) ! if (Mode == OMrunning && !HISwrite(&Data, "")) syslog(L_ERROR, "%s cant write history %s %m", LogName, Data.MessageID); #endif /* defined(DO_REMEMBER_TRASH) */ --- 1815,1821 ---- syslog(L_NOTICE, "rejecting[perl] %s %s", HDR(_message_id), buff); ARTlog(&Data, ART_REJECT, buff); #if defined(DO_REMEMBER_TRASH) ! if (Data.MessageID && Mode == OMrunning && !HISwrite(&Data, "")) syslog(L_ERROR, "%s cant write history %s %m", LogName, Data.MessageID); #endif /* defined(DO_REMEMBER_TRASH) */ *************** *** 1843,1849 **** buff); ARTlog(&Data, ART_REJECT, buff); #if defined(DO_REMEMBER_TRASH) ! if (Mode == OMrunning && !HISwrite(&Data, "")) syslog(L_ERROR, "%s cant write history %s %m", LogName, Data.MessageID); #endif /* defined(DO_REMEMBER_TRASH) */ --- 1859,1865 ---- buff); ARTlog(&Data, ART_REJECT, buff); #if defined(DO_REMEMBER_TRASH) ! if (Data.MessageID && Mode == OMrunning && !HISwrite(&Data, "")) syslog(L_ERROR, "%s cant write history %s %m", LogName, Data.MessageID); #endif /* defined(DO_REMEMBER_TRASH) */ -- J. Porter Clark, d/b/a +--+ |oo| The Unknown News Administrator | | news@news.msfc.nasa.gov `^^'