> On Oct. 22, 2013, 1:05 a.m., Benjamin Hindman wrote: > > src/logging/logging.cpp, lines 104-106 > > <https://reviews.apache.org/r/14800/diff/1/?file=368956#file368956line104> > > > > I think we want to add an else if which turns logging to stderr back > > off: > > > > if (!flags.quiet) { > > FLAGS_stderrthreshold = 0; // INFO. > > } else { > > FLAGS_logtostderr = false; > > } > > > > Otherwise, if someone specifies 'quiet' they'll still get logging to > > stderr. Make sense? > > Yuval Pavel Zholkover wrote: > Sorry, I'm learning to use reviewboard properly. > FLAGS_logtostderr must always be set true whenever log_dir is empty. It > actually prevents glog from writing files to /tmp. > > Current mesos behaviour is as follows: > * no extra arguments - logs with severity level INFO and higher are > written to stderr and to files in /tmp (this is wrong). > * without log_dir and --quiet - logs with severity level ERROR and higher > are written to stderr and to files in /tmp (this is wrong). > * with log_dir set - logs with severity level INFO and higher are > written to stderr and to files in <logdir>. > * with log_dir set and --quiet - logs with severity level ERROR and > higher are written to stderr and to files in <logdir> > > Propsed patch v2 behavour is as follows: > * no extra arguments - logs with severity level INFO and higher are > written to stderr and no files are outputed. > * without log_dir and --quiet - logs with severity level INFO and higher > (this is wrong) are written to stderr and no files are outputed. > * with log_dir set - logs with severity level INFO and higher are > written to stderr and to files in <logdir>. > * with log_dir set and --quiet - logs with severity level ERROR and > higher are written to stderr and to files in <logdir> > > I think the patch v2 behaviour makes more sense, as the only thing it > does is being extra verbose when log_dir is not set and --quiet is requsted. > > Ben Mahler wrote: > Hey Yuval, is there an issue in the glog issue tracker related to the bug > you described below? > > Yuval Pavel Zholkover wrote: > Hi Ben, thanks for taking the time to review this. I did some digging and > I think this is related but not the exact issue: > http://code.google.com/p/google-glog/issues/detail?id=128 > >
Perhaps to get around the extra verbosity we can do: google::SetStderrLogging(3); As suggested in the linked issue? - Ben ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/14800/#review27268 ----------------------------------------------------------- On Oct. 22, 2013, 6:28 a.m., Yuval Pavel Zholkover wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/14800/ > ----------------------------------------------------------- > > (Updated Oct. 22, 2013, 6:28 a.m.) > > > Review request for mesos, Benjamin Hindman, Ben Mahler, and Vinod Kone. > > > Repository: mesos-git > > > Description > ------- > > The current intialization of glog creates log files in /tmp even when the > log_dir command argument is not passed. > Setting FLAGS_logtostderr to true disables all log file creation/writes. > > > Diffs > ----- > > src/logging/logging.cpp 850fb3c > > Diff: https://reviews.apache.org/r/14800/diff/ > > > Testing > ------- > > manual - during make check /tmp gets filled with glog logs (timestamped and > symlinked) > the patch fixes it > > > Thanks, > > Yuval Pavel Zholkover > >
