Re: RFS: bandwidthd - tracks network utilization per ip and draws graphs.
Replying to my own mail yesterday....
David: Please don't merge the patches I've sent you.
I've looked into the "skipping current run" problem this morning.
I haven't found the root cause but here are my conclusions so far...
bandwidthd-1.2.1b-13 (debian package)
bandwidthd-1.2.1b + all patches sent upstream
bandwidthd-1.2.1b + fork patch sent upstream
(?) bandwidthd-1.2.1b (plain upstream)
CommitData sets MayGraph=FALSE before initiating a graphing run to
prevent another graphing run before the first one is finished and then
calls WriteOutWebpages. MayGraph=TRUE is set when there is a (grapher)
child to reap.
WriteOutWebpages doesn't fork a child if there's nothing in DataStore.
One solution to this would be to change WriteOutWebpages to return an
error code so CommitData can reenable MayGraph if WriteOutWebpages
fails to not prevent future graphing runs forever.
Alternative solution: Check the datastore before changing MayGraph and
calling WriteOutWebpages (this OTOH can't handle fork failures).
The "skipping current run" problem is there because the minor bug above.
The real bug is somewhere else though and I need to find out why the
DataStore is empty.
I can't see how any of my changes can cause this problem. I will have to
look closer on the fork patch and investigate this further....
If the workarounds mentioned doesn't cure it for you or have anything to
add in tracking down the problem please drop me a mail and I'll look at
it as soon as I get my next chance to investigate.
I'll post a status update later on....
Thanks to everyone testing the package!
Thanks Adeodato for the suggestion on "lists @lists.debian.org" in
muttrc which I've now added. Still I want to urge everyone to please CC