Hi Ken,
----- Original Message -----
> A few more to report ... all using the Makepkgs, install, new daemons
> running, QA check -g sanity (354 and possibly 184 failing) recipe.
>
I think 354 is fallout from your earlier pmRecordSetup change in commit
3ecf28f33c (its failing here too, sorry been meant to follow up on this
earlier). If I revert that commit, problem goes away ... it looks like
that commit may be inadvertently removing the actual folio file?
A QA test update like below gives more clues (MKTEMP is now empty since
the file doesn't exist, and the filtering breaks).
diff --git a/qa/354 b/qa/354
index 91f2ceb..3fc0251 100755
--- a/qa/354
+++ b/qa/354
@@ -109,6 +109,7 @@ echo
$here/src/record foo no-1 1 $host localhost $host localhost 2>&1 >$tmp.out &
sleep 10
MKTEMP=`ls -d * | grep -v foo | grep -v '\.'`
+echo MKTEMP is \"$MKTEMP\" >> $here/$seq.full
_build_sed $tmp.out
_filter $tmp.out
cat $tmp.out >>$here/$seq.full
I'll push that diagnostic change in permanently.
> FreeBSD 8.2-RELEASE-p9 amd64 - after some build-related hacking, built a
> tarball which installed cleanly ... pmcd started but not much success
Ah, I remembered I have access to a FreeBSD 9.1-PRERELEASE machine from
a friend (but not root access) - I've fixed up a handful of build issues
there that may help here.
> beyond this, I think because of running as non-root issues that have
> never been sorted out for FreeBSD (and indeed for any tarball based
> packaging of PCP).
Ideally we'd figure out proper packaging for FreeBSD and automate the new
user creation (and namespace root Rebuild, and so on) I guess. But we'd
need an interested/engaged BSD hacker there or someone with time to spare
to make it work (all of which rule me out!).
cheers.
--
Nathan
|