It seems like setting the LISTENER_EXECUTOR_THREAD_COUNT to value 1 instead of 2 reduces probability of the issue to happen.
On Mon, Dec 5, 2011 at 6:16 PM, Yuri Z <vega...@gmail.com> wrote: > Yes, this is bug is annoying. I mean the "No conversations in > this wave". The most nasty part is that I cannot reproduce it in my dev > environment as well. This is not a new issue though, it happened also > before, but at lower frequency. Server restart fixes it. > My theory is that the "No conversations" bug is related to the "duplicate > delta" warnings in the log. I.e. the first delta in the wavelet, specifies > the author - it's like - whoever writes the first delta to the wavelet - is > the author. But then the first delta is somehow recorded twice. And then on > the wavelet open request - it somehow reads incorrect delta and determines > that the participant is not allowed to access is, so it sends an empty > snapshot. Which causes the "No conversations" issue. > > > On Mon, Dec 5, 2011 at 5:57 PM, Vicente J. Ruiz Jurado < > v...@ourproject.org> wrote: > >> Hi there: >> >> Since some weeks ago we are experimenting some exception randomly: >> https://issues.apache.org/jira/browse/WAVE-307 >> (see RemoteWaveViewService.getServerVersion) that makes WIAB very >> unstable/unusable because we cannot edit or reply some waves, or we lost >> editions, or access to that wave with the message "No conversations in >> this wave". >> >> I'm very lost looking in the code and logs, also because I cannot >> reproduce it in my development workspace. >> >> Can anybody more experienced help on this, giving some info, advice, >> reasons for that exception, etc? Maybe someone is looking already on it... >> >> Thanks, >> >> Vicente >> > >
