erict@flatline.UUCP (J. Eric Townsend) (04/06/89)
The replies I recieved up until Wed Apr 5 9:31 pm are included at the bottom of this message. In article <654@icus.islp.ny.us> lenny@icus.islp.ny.us (Lenny Tropiano) writes: >|>Well, here's the error: >|>pathalias: out of memory (2460k allocated); notify the authorities >|>I have this sneaking suspicion that I can only allocate 2460k per >|>process under my flavor of unix. (3b1 Unix 3.0, kinda like Sys V rel 0.0). >This is only 1480k, not 2460k? What are you trying to pathalias? >Granted I have 3.5MB in my machine ... that might have something to do >with it ;-) I thought the only limit on this virtual memory machine was I have 3.5Mb, 67Mb HD. From what I understand, there's a ~2.3Mb per process limit. I thought there was memory until the swap space ran out. Oh well. du -s /usr/lib/mail/maps 5506 I guess that's a lot, huh? Maybe there's some sort of compile flag for when I compile pathalias? I don't have the man page, so I'm at a slight disadvantage... Lenny: If you know of any sort of information I could give you that would help in coming up with at least a description of what's happening I'd be glad to dig through my system. ============================================================ >From att!skep2!wcs Mon Apr 3 20:39:45 1989 remote from texbell well, there are several possibilities. The first is that you're hitting ulimit (ulimit is in the System V manual under "braindamage"). Since you're root, you can set your ulimit large before running pathalias. There are some related process-memory limits in the kernel that can't be avoided as easily. Another is that your machine really has limits on how much memory it can allocate. It may be limited by physical memory - I don't remember what the 3B1 does about virtual memory? I don't know if pathalias keeps stuff in arrays, or if it just uses pointers to tie things together - while the system doesn't use segments, int's are 16-bits. This means that an array can only have 32767 entries in it. ============================================================ From: David Herron -- One of the vertebrae <rutgers!ms.uky.edu!david> yes you can only allocate ~2.5meg of memory on 3b1's look around and you might find the disk based version of compress for 16-bit machines. then send a letter bomb to convergent and/or att ============================================================ From: uunet!research.att.com!wilber%alice I'm afraid ~2.5 Mb is the limit on user space for any single process on the 3b1. (An MMU limitation.) Looks like you have some serious hacking to do on pathalias. ============================================================ From: uunet!berner!richard (Richard Greenall) There probably isn't a whole lot you can do about it. I run pathalias here without problems on a xenix machine, but that is because I have 10M of memory. The only thing that you can probably do is to run it on a subset of the paths, and get another site (I find universities are good) to be a smart-host for you. ============================================================ From: rutgers!psuvax1.cs.psu.edu!pitt!investor!rbp I used to run Sys III on an Altos box and couldn't do pathalias. I suspect you just don't have enough physical RAM, period. If you can't get more RAM, try shutting down to singleuser and running it there. With the reduced overhead, you might just make it. I currently run on a 4Mb box under SysV. Even herre I get nasty messages about not being able to get the core requested. Fortunately, on this system it just sits in swap space and keeps trying until it can get what it needs. -- J. Eric Townsend | "Type 'make'. If you get errors, stop and fix them." uunet!sugar!flatline!erict | -- Eric S. Raymond, in TMNN documentation. bellcore!texbell!/ 511 Parker #2 |EastEnders Mailing List: Inet: cosc5fa@george.uh.edu Houston,Tx,77007 |eastender@flatline.UUCP