rdc30@nmrdc1.nmrdc.nnmc.navy.mil (LCDR Michael E. Dobson) (05/31/91)
Last week I posted that the listen process for port 25 was dying for uknown reasons using sendmail-5.65b+IDA-1.4.3. I still don't know exactly why this was happening, especially since ps -ef showed the daemon was still active, just that netstat -a no longer showed a *.smtp LISTEN process. I switched back to the vanilla 5.59 that came with WIN/TCP 3.0.1 on this machine and it worked fine but I needed the pathalias stuff. When comparing the .cf files for the two versions, I noted that the load average (OX) parameter was 15 for the 5.59 version, but the .m4 file I used to create the 5.65b version used a value of 7. Increasing the value to 15 for 5.65b+IDA seems to allow it to work. Hope this can be of use to others who may have experienced similar troubles. -- Mike Dobson, Sys Admin for | Internet: rdc30@nmrdc1.nmrdc.nnmc.navy.mil nmrdc1.nmrdc.nnmc.navy.mil | UUCP: ...uunet!mimsy!nmrdc1!rdc30 AT&T 3B2/600G Sys V R 3.2.2 | BITNET: dobson@usuhsb or nrd0mxd@vmnmdsc WIN/TCP for 3B2 | MCI-Mail: 377-2719 or 0003772719@mcimail.com
paul@uxc.cso.uiuc.edu (Paul Pomes - UofIllinois CSO) (06/01/91)
rdc30@nmrdc1.nmrdc.nnmc.navy.mil (LCDR Michael E. Dobson) writes: > When comparing the .cf >files for the two versions, I noted that the load average (OX) parameter was >15 for the 5.59 version, but the .m4 file I used to create the 5.65b version >used a value of 7. Increasing the value to 15 for 5.65b+IDA seems to allow >it to work. Does the displayed load average on your machine as shown by uptime agree with what sendmail calculates using getloadavg()? The latter can be determined by using the -d3.1 flag, e.g., sendmail -bt -d3.1. If you normally have a busy machine, then the distributed values for Ox and OX will need tweaking. For instance our Convex C240 uses 20 for queueing and 40 for rejecting connections. Compiling with SETPROCTITLE defined will show what sendmail is up to when checking with ps. The getloadavg.c file has become the hairest portion of the sendmail code. Unfortunately kernel diving is never neat. /pbp -- Paul Pomes, Computing Services Office University of Illinois - Urbana Email to Paul-Pomes@uiuc.edu