Global connection timeout is 30 sec
Delay between connection retries is 20 sec
Since it's hard to get this to occur at my end (I have to wait for a missing article 'to show up'), which way do you suggest changing these two parameters and by how much? Should I double, treble, halve, quarter them?
FWIW: I have doubled both settings and am now waiting for a 'missing article'.
Thank you for any help you can give me in pinpointing this 'possible bug'.
update 09.02.07-
the doubled numbers (60 sec TO, 40 sec delay) ended up with all 8 threads trying to connect to my second backup server (which still doesn't want to respond, but then that's why I got a Ngroups account). I have now halved the numbers (15 sec TO and 10 sec delay) and am waiting for an unavailable article to show up. So far it looks like doubling made it worse (as all threads end up in a useless state) and halving is still up for 'review'. Just to make things clear- my complaint is not with the 'failure' to use a server, but with the 'failure' to stop trying to use an unavailable server. My feeling is that after N failed attempts to connect to a backup server, the article should be flagged as 'not available' on that server and AltBinz should move on to the next server (if there is one) or give up on the article as it does if there are no backup servers configured.
update 09.02.10-
the halved numbers (15 sec TO, 10 sec delay) also ended up with threads trying to connect to my second backup server (although I only found 4 of the 8 threads 'hung'). I have returned the timings to the default values and removed the second backup server from the configuration (administrative solutions are almost always faster to implement than technical solutions). Thanks for looking into this.
