[volunteergrid2-l] policy about clients

[Marco Tedaldi]

Am 10.09.2012 21:55, schrieb Ruben Pollan:
> Quoting erpo41 at gmail.com (2012-09-10 20:41:44)
>> I have the same issue with unhappiness errors. My solution was a shell script;
>> I can post it if you want.
>
> Right now I'm doing something like: 'while 1; do tahoe backup ...; done'
> But this one will keep trying even if the backup finish. How do you check if it
> finish?
>
I have had the same problem.
I've realized, that it is most dominant if shares happy == shares total
The further apart (shares total > shares happy) tgese are, the better it 
works.

So set the shares total to the total number of available nodes and make 
shares happy something else.
I gut the impression, that tahoe backup selects a set uf nodes when the 
command is started and does not re-select for every file. And if a node 
is unavailable once, it is thrown out of the selection. This way, the 
list af available nodes for the command gets smaller and smaller until 
it is smaller than "shares happy" which leads to the error.
When tahoe backup is called again, a new list will be generated...

so, just waste space for some extra shares and tahoe backup will work.

Oh, and don't try to backup all the file at once. You can just start 
with one sub-directory at once and than "climb down the tree".

If you have backed up all sub directories you can just backup the 
parent. Tahe will see, that the files and sub dirs are all already there 
and just create the directory now which is quite fast.

best regards

Marco