Bugzilla Bug 203646

If rpc.statd (from nfs-utils-1.1.0-r1) is not running, and I try to start it
with "/etc/init.d/rpc.statd start" nothing happens, i.e. there is no output and
statd is not started.

I've traced the problem to the is_running() function. The single command in
that function should be returning nonzero if rpc.statd is not running, but this
does not happen. I've checked by changing the function to:

is_running() {
   killall -0 rpc.statd 2>/dev/null
   echo $?
}

With that change, if rpc.statd is not running, "/etc/init.d/rpc.statd start"
prints 0, and nothing else happens (in this case it would be expected, since
the echo command itself will return 0, but the point is that killall does,
too).

The strange thing is, if I run the command outside emerge (if rpc.statd is not
running), I get the expected behaviour:

# killall -0 rpc.statd; echo $?
rpc.statd: no process killed
1

I'm confused. Does portage perhaps override killall to an internal function? In
any case, the behavior is wrong..

Reproducible: Always

Steps to Reproduce:

*** This bug has been marked as a duplicate of bug 60362 ***

I did find that bug before, along with a number of others, but nothing there
helps solve this problem. Please don't close bugs without looking into them.

Yeah I did look at the bug. Complain to whomever closed the original one, this
is the same dupe like the rest of others there.


*** This bug has been marked as a duplicate of bug 60362 ***

It does actually contain more info than the others. And since I can't reopen
them, what am I supposed to do? Mail vapier (he closed the "original" bug)?
If you were trying to be helpful, you might have assigned the bug to him insted
of closing it.

This is exact same thing like Bug 175287 which was marked as dupe bug 60362; so
yeah, take this w/ vapier and ask him why's the original bug resolved.

i dont know why you think it's a duplicate because it isnt

prob because i mis-duped Bug 175287 ... oh well

legate: do you have RPCSTATDOPTS set ?

no, that isnt the problem ... problem is that the killall detects the init.d
script itself which is also named "rpc.statd" :/

should be fixed in cvs

http://sources.gentoo.org/net-fs/nfs-utils/files/rpc.statd.initd?r1=1.3&r2=1.4

Wow, way to think outside the box.. It's one of those things I never would have
come up with, without going all-out with ltrace or the like. Especially since
the next call in start() is a "return 0", which I just assumed was getting run.

And I was really starting to question my sanity.. Whatever you're making, you
need to be in a higher pay grade. Thanks.