orte_data_server: fix a typo in ORTE_PMIX_PUBLISH_CMD handling

[ggouaillardet]

correctly balance some parenthesis ...

Fixes #4153

Thanks Austen Lauria for the report

This is a one-off commit for the v3.0.x branch, master was fixed as part of a larger commit,
and the v2 branches are unaffected.

Signed-off-by: Gilles Gouaillardet [email protected]

[ggouaillardet]

@bwbarrett this fix is obvious (though finding the root cause was not ...) and though the faulty path is not always taken and does not always cause a crash, i made this PR a blocker so hopefully it can make it for v3.0.0.

[rhc54]

Thanks!

[rhc54]

The failure is in the opal_path_nfs test, which has nothing whatever to do with this PR.

[gpaulsen]

This is important to get into v3.0.0.

[bwbarrett]

bot:ompi:retest

[bwbarrett]

It looks like something changed on the Cray so that the nfs check now is consistently failing. I'm sure it doesn't have anything to do with this patch, but it needs to be addressed. @hppritcha, can you take a look? The test is only faling on the Cray and is consistently failing there.

[hppritcha]

will do. We can ignore for this PR.

[hppritcha]

actually there's nothing screwed up on the crays.

problem is that the entries in /proc/mounts on some cray systems can be enormous, much larger than the buffer being used in the get_mounts method in opal_patn_nfs.c. Anyway, one is suppose to use getmntent and friends to avoid this problem.

Do we want to fix the test or remove it from make check?

[jjhursey]

We should resolve the CI issues separate from this PR. This PR is an obvious, correct, critical fix. We should get it in the branch soon (today?) - it's absence is causing testing problems with optimized builds the longer we let it sit here.

[jsquyres]

@jjhursey #4179 was just merged that should fix the issue. Let's see if we can get a nice green checkmark now...

bot:ompi:retest

[rhc54]

Urrr...that commit went into master. How is that going to fix this PR into v3.0?

[bwbarrett]

@jjhursey, the cranky RM for 3.0 (ie me) has said no green, no merge. So we need to get CI fixed...

[jsquyres]

@rhc54 Errr... right. Duh. v3.0.x PR filed here: #4180

PRs for other branches to be filed shortly.

[jjhursey]

Looks green to me 😄