Date: Sun, 19 Jan 1997 09:12:44 -0800 (PST) From: Bill Paul <wpaul> To: mark@grondar.za (Mark Murray) Cc: current@freebsd.org, peter@freebsd.org Subject: Re: NIS breakage Message-ID: <199701191712.JAA09406@freefall.freebsd.org> In-Reply-To: <199701191608.SAA04321@grackle.grondar.za> from "Mark Murray" at Jan 19, 97 06:08:00 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > Hi > > I have a 386sx/40 with IDE, 8MB of RAM. I have a tun0 (currently unused) > ed0 [my home ethernet backbone with a subnet of my class C], lo0 [127.0.0.1] > and lo1 [127.0.0.2 and another subnet]. I have Apache attached to one of > the IPs on lo1. I wish to heck I knew what it is you're trying to accomplish with this second loopback interface. > My system is 3.0-current (19th JAN). The problem may have started a week > ago, but I only noticed it about 3 days ago, and I waited for this > "make world" and install before reporting it. The "make world" did not > help or change anything AFAIK. > At times, it is difficult to get sendmail to answer (only difficult, not > impossible like before - it just takes a LOOONG time). > It also logs these: > Jan 19 17:25:27 grunt sendmail[166]: NOQUEUE: SYSERR(root): getrequests: accept: Bad address > > NIS is _horribly_ fragile - If I change passwords, the following happens > (logged to syslog): > > Jan 19 16:57:28 grunt portmap[161]: svc_run: - select failed: Bad address > Jan 19 16:57:28 grunt portmap[161]: svc_run returned unexpectedly > Jan 19 16:57:28 grunt /kernel: pid 161 (portmap), uid 1: exited on signal 6 > > On the client side, everything just hangs - ^C needed to break out. > > Just before attempting to change the password, I did a `ps -ax', and > everything looked normal - no multiple yp* or anything. Uh, Mark? It said that portmap died. What did you expect was going to happen? Nothing in RPC works without portmap. > I get lots of these: > > Jan 19 16:58:26 grunt ypbind[174]: select: Bad address > > ...but as I have been getting them for a while, I don't think they hurt > too much. > > Clues? portmap and ypbind use their own svc_run() loops. (ypserv does too, and it's a little odd that you're not having trouble with that, unless the server is running on a different host.) I'm confused, mainly because sendmail blew up in an accept() rather than a select(). I was thinking maybe something in the new RPC changes Peter made might be clobbering file descriptors, but now I'm not too sure. Now for the questions: do you have any other systems running the same build of -current? If so, do they have the same problems? Can you also do me a favor and get rid of that second loopback interface and see what happens then? (And I mean really get rid of it: configure it out of the kernel, don't just ifconfig it down.) The only way I'm going to be able to track this down is to load a -current snapshot on my test box, since I don't see problems like this with 2.2-BETA (thank goodness). *sigh* I don't have time for this. -Bill
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199701191712.JAA09406>