Just observed a situation where the current polling code breaks in case of input being the exact multiple of the read buffer size (4096). The cardinal issue is that the code didn't expect POLLHUP to appear when not original asked for capture, and in turn, not expecting poll will return positive number in such case, leading to fallback to "nothing read of malformatted input" condition. This is partly in one line with previously discovered issues ([bug 732483]), which led me to some more code base reviews, resulting even in a patch to this very issue (although it was more like fix "suspicious" than "apparent" back then): [1]. Unfortunately there was no direct justification for that patch at that time. [1] https://git.fedorahosted.org/cgit/conga.git/commit/?id=631b17aab8f44396c4cf373bc4466bbfcaab6563
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux release for currently deployed products. This request is not yet committed for inclusion in a release.
Note that the size relates to the overall <request/> part of the RICCIAPI, which is being passed to the particular module, not to the size of cluster.conf. Although in the initial observation, it was a set_cluster.conf request with cluster.conf size roughly 7580 B (exact equation for triggering size shouldn't be too hard to figure out).
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHSA-2014-1194.html