Description of problem: perl-Net-Server-0.93-1.fc5 breaks amavisd-new. mailq shows: (host 127.0.0.1[127.0.0.1] said: 451 4.5.0 Error in processing, id=08431-01, parts_decode_ext FAILED: parsing file(1) results - missing last 1 results at (eval 41) line 154. (in reply to end of DATA command)) from /var/log/maillog: Apr 26 00:02:01 myhost amavis[8525]: (08431-01) run_command: child process [8525]: Error closing main::stdin: \344\270\215\346\255\243\343\201\252\343 \203\225\343\202\241\343\202\244\343\203\253\350\250\230\350\277\260\345\255 \220\343\201\247\343\201\231 at /usr/sbin/amavisd line 1872.\n Apr 26 00:02:01 myhost amavis[8431]: (08431-01) TROUBLE in check_mail: parts_decode_ext FAILED: parsing file(1) results - missing last 1 results at (eval 41) line 154. Apr 26 00:02:01 myhost amavis[8431]: (08431-01) PRESERVING EVIDENCE in /var/spool/amavisd/tmp/amavis-20060426T000201-08431 Apr 26 00:02:01 myhost postfix/smtp[8524]: 0896133167: to=<morioka.com>, orig_to=<morioka>, relay=127.0.0.1[127.0.0.1], delay=1, status=deferred (host 127.0.0.1[127.0.0.1] said: 451 4.5.0 Error in processing, id=08431-01, parts_decode_ext FAILED: parsing file(1) results - missing last 1 results at (eval 41) line 154. (in reply to end of DATA command)) abobe "\344\270\215..." is UTF-8 message in japanese â䏿£ãªãã¡ã¤ã«è¨è¿°åã§ã". This means "Bad file descriptor". I downgraded to old perl-Net-Server-0.90-2.fc5, and amavisd-new works again. Version-Release number of selected component (if applicable): perl-5.8.8-4 perl-Net-Server-0.93-1.fc5 amavisd-new-2.3.3-5.fc5 postfix-2.2.8-1.2 How reproducible: always Steps to Reproduce: 1.Setup Postfix and Amavisd-new. 2.Send a mail 3.mailq Actual results: mail is defferd by error. Expected results: mail is sent. Additional info:
Sorry, this is already solved in upstream by amavisd-new-2.4.0. http://www.ijs.si/software/amavisd/release-notes.txt
It seems thr Net::Server changed something to fix a boog amavisd-new worked around, and as a result amavisd-new versions which include the workaround are broken now : http://rt.cpan.org/Public/Bug/Display.html?id=18402 Upstream advice is to switch to amavisd-new 2.4.0 -> reassigning (Sorry I didn't notice this before releasing, all tests where green. But now upgrading amavisd is probably the best solution)
Steve, what do you think ?
Upgrading amavisd-new is definitely the right choice. I just updated devel and requested a build. I want to do some testing with that before pushing out an update for FC-4/FC-5.
btw, this can wait some time but I'd really like to remove the zoo dep (and nuke zoo from extras) and use the sqlite backend instead of the bdb one by default (bdb is always breaking on upgrades and the security flyby of zoo after the last vuln was not encouraging)
It looks like the new version of amavisd-new is going to be a painless upgrade, so I'm pushing builds for FC-4, FC-5, and devel. In response to comment #5: > btw, this can wait some time but I'd really like to remove the zoo dep (and nuke > zoo from extras) and use the sqlite backend instead of the bdb one by default Removing the zoo dep is easy. Are you aware of any of the other hard dependencies being risky? Got a patch to switch to sqlite by default? I'd be completely OK with that.
zoo is the worst of the lot. The others have either a cleaner/more recent codebase or are still maintained (speaking from memory, not taking the livna stuff into account) As for sqlite I didn't look at it a lot, just noticed in the startup messages it could be used as backend (and the bdb backend is breaking every month in rawhide). I'll try to look at it some more. Thanks for the quick update and sorry for my Net::Server messup