I'm in the same boat. When this happens, I notice that Send-Q is always stuck at a fixed nonzero value for one or more connections between my node and someone else's. To see the Send-Q, run netstat -ep.
Thanks, Eric On Jul 1, 2012 7:51 AM, "Marco Tedaldi" <[email protected]> wrote: > Hello everyone > > While using tahoe backup to make a backup of my digital pictures I came > across a strange behavior. > > I'm using tahoe 1.8.3 > > When trying to upload directories with subdirectories and a lot of files > in them, thaoe fails with happiness errors. > > It tells me something like: Want to upload to 12 server so data can be > recovered from 7 servers but could only upload to 11 servers. > > This despite the fact that there are 18 server up and accepting shares. > Restarting the job sometimes helped. > Starting another job always worked > Restarting my local tahoe node always helped > > So I get the impression that tahoe initially selects a set of "total > servers" from all the servers when "tahoe backup" is started. But than it > seems to stick with these servers. If some of these selected servers fail > to accept shares, tahoe does not seem to select other servers instead > (except if tahoe node is restarted) > > If total server is way higher than happy, the problem does not occur that > often. > If total = happy the problem occurs very often! > > > I have been able to work around that problem by setting total servers a > bit higher than happy and uploading data in smaller chunks (so a directory > does not take more than a day or two to finish). > > > Is this like tahoe works? > Is it a bug? > > best regards > > Marco > ______________________________**_________________ > tahoe-dev mailing list > [email protected] > https://tahoe-lafs.org/cgi-**bin/mailman/listinfo/tahoe-dev<https://tahoe-lafs.org/cgi-bin/mailman/listinfo/tahoe-dev> >
_______________________________________________ tahoe-dev mailing list [email protected] https://tahoe-lafs.org/cgi-bin/mailman/listinfo/tahoe-dev
