Re: [CDBI] C::DBI Memory usage
"Aaron Trevena" <[email protected]>
| Newsgroups | gmane.comp.lang.perl.modules.class-dbi |
|---|---|
| Message-ID | <[email protected]> |
On 24/05/06, Matt S Trout <dbix-class-uY5y5/[email protected]> wrote: > > I'm working on scheduling daemon in perl, it uses C::DBI to handle > > DB/Object Persistence, etc as you'd expect. > > These days I'd expect DBIC or RDBO, but hey, if you enjoy the legacy > thang :) Meow! The time to learn a new ORM to the level required isn't available, CDBI has JustWorked(TM) so far. Have you any gurantee your trendy new ORMs are likely to prevent this problem - I'm guessing you haven't done any memory benchmarking over several days. > First off, try clearing the LiveObjectIndex every so often. It's almost > certainly not the culprit but it's so easy to try you might as well. I have set the weaken is unavailable flag which should prevent use of the LiveObjectIndex. > Second, get yourself a copy of Test::Memory::Cycle (and currently > Devel::Cycle 1.04 since 05 and 06 have a slight issue - or look at the > DBIx::Class list for how to tweak a more recent one to work happily) and > start lobbing cycle tests at everything in sight. I tried this and it wasn't at all helpful - I'll give some of the older ones a go. Whenever I've tried to watch the memory usage of individual variables I've been overloaded with hundreds of them unable to see clearly what is increasing or decreasing - let alone patterns responsable for memory usage over time. > > Also has anybody else encountered this problem, or succesfully used > > CDBI in a long-running ( i.e. over 48 hours ) program? > > Yes, but it was back around 0.96 and I don't remember if we ever > properly leak-tested it. drat. A.