Sendmail rset command hangs socket on 1.6-Release

Matthew Dillon dillon at apollo.backplane.com
Tue Aug 1 18:28:29 PDT 2006


:We have a cluster of mail servers, one of which we have upgraded to
:1.6-Release (from 1.2) and then subsequently did a fresh install of 1.6
:on after the issue first arose.
:
:In essence, using the supplied sendmail and having the connected
:sendmail session issue a rset command results in that sendmail process
:hanging (only by issuing the telnet ^] command to get to a telnet prompt
:and issuing quit can the session be killed). If instead of an rset
:command the quit command is issued, the session will end (i.e. the
:telnet session terminates) but the sendmail process with "quit" appears
:in the process list and will not go away unless killed.
:
:Has anyone seen this odd behavior? At first I thought it was due to a
:milter, so I removed it from the .mc file and rebuilt sendmail.cf but no
:change in socket behavior. How can I help in terms of troubleshooting?
:
:Sven
    
    That is really odd. I wonder if it is related to the posix issues
    that were reported recently.

    If you can repeat it easily with telnet, could you use tcpdump
    and ktrace (on the forked sendmail demon after you connect) to
    figure out what happens when you try to send it the 'rset' ?
    With tcpdump it should be possible to determine if the 'rset' gets
    transmitted, and with ktrace it should be possible to determine
    if sendmail sees it.

					-Matt
					Matthew Dillon 
					<dillon at xxxxxxxxxxxxx>





More information about the Bugs mailing list