> On Jan. 22, 2014, 8:33 p.m., Mike Drob wrote: > > src/server/src/main/java/org/apache/accumulo/server/trace/TraceServer.java, > > line 233 > > <https://reviews.apache.org/r/17129/diff/4/?file=435551#file435551line233> > > > > I don't understand what happens to the rest of the code path if we set > > a null writer.
The flush and receiver threads will get a null writer when they next look, which will cause them to gracefully avoid writes (logging a warn in the case of the receiver). the behavior of the reset thread depends on what the previous value in the atomic reference was. if it was non-null, we'll try to close. if it was null, we'll skip. - Sean ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/17129/#review32531 ----------------------------------------------------------- On Jan. 22, 2014, 8:25 p.m., Sean Busbey wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/17129/ > ----------------------------------------------------------- > > (Updated Jan. 22, 2014, 8:25 p.m.) > > > Review request for accumulo and Eric Newton. > > > Bugs: ACCUMULO-2213 > https://issues.apache.org/jira/browse/ACCUMULO-2213 > > > Repository: accumulo > > > Description > ------- > > ACCUMULO-2213 Make writer recovery in the Tracer more robust. > > * Cleans up writer reseting in the TraceServer, avoids overly broad > catching. > * tones down log levels in TraceServer to WARN because trace information > is transient and we retry everything. > > > Diffs > ----- > > src/server/src/main/java/org/apache/accumulo/server/trace/TraceServer.java > 4d89e9c13f89e71e674875db32cf267e045d6131 > > Diff: https://reviews.apache.org/r/17129/diff/ > > > Testing > ------- > > running on cluster with injected failures now > > > Thanks, > > Sean Busbey > >
