----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/28218/#review65412 -----------------------------------------------------------
src/log/log.cpp <https://reviews.apache.org/r/28218/#comment108524> This log is missleading. It's LogProcess recovery fails (not reader process). Can you instead add this logging in LogProcess? (i.e., in ```LogProcess::_recover()```) src/log/log.cpp <https://reviews.apache.org/r/28218/#comment108527> Ditto. Do the logging in LogProcess. src/log/log.cpp <https://reviews.apache.org/r/28218/#comment108528> Add a blank line after VLOG. HEre and everywhere else. Can you instead add this logging in Replica process because essentially this is calling replica->read(from, to). src/log/log.cpp <https://reviews.apache.org/r/28218/#comment108529> I don't think this is necessary. Mind killing it? src/log/recover.cpp <https://reviews.apache.org/r/28218/#comment108530> Can you move this log line into 'finished'? src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108532> Remove the 'process' suffix? src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108533> Looks like this is uncesseary given that all the subsequent code path has a VLOG. src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108535> This is not precise. ``` VLOG(2) << "Writer failed to get elected, retrying"; ``` src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108536> VLOG(2) << "Writer got elected at position " << position.get() << ", index is " << index.get(); src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108541> Kill this. src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108542> Kill this. src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108545> "Applying Operations (" << entries.size() << " entries)" src/state/log.cpp <https://reviews.apache.org/r/28218/#comment108546> Kill this one. - Jie Yu On Nov. 19, 2014, 9:58 p.m., Timothy Chen wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/28218/ > ----------------------------------------------------------- > > (Updated Nov. 19, 2014, 9:58 p.m.) > > > Review request for mesos, Ben Mahler, Jie Yu, and Niklas Nielsen. > > > Repository: mesos-git > > > Description > ------- > > There has a quite few cases where we see a timed out fetch when master is > first launched, and I think there isn't enough logging to shed some lights > besides the failed to fetch after x minutes log line. > This rb adds more logging into replicated log recovery. > Not meant for merging yet, posting here for reviews and discuss places we > like to add more logging. > Review: https://reviews.apache.org/r/28218 > > > Diffs > ----- > > src/log/log.cpp b3c6c20d734ea82d8531217d53decf1707183565 > src/log/recover.cpp 26140654fda5011e89d0081e916a88c9681cd783 > src/state/log.cpp c0c81d551f465534daeaf03f18c0baa4c19754e6 > > Diff: https://reviews.apache.org/r/28218/diff/ > > > Testing > ------- > > make > > > Thanks, > > Timothy Chen > >
