I've been looking into various memory leak issues over the last week, with some limited success (see #2824 patch).
Through various tests and from looking at data from users I'm suspecting that this is only a partial solution, and that there may be other leaks, though I've as yet been unable to characterize them sufficiently. Here's what I suspect: * puppetd will, under certain circumstances retain an increasing volume of data from factor through a series of runs; this eventually settles down to a steady state, * objects in the report hierarchy sometimes accumulate * an error while writing the cached copy of the catalog can leave large SYCK structures in memory for an indefinite amount of time (through multiple runs). * network timeouts can leave orphaned structures, but these appear to be released for collection on the next subsequent successful run * recursing through a large directory structure (e.g. a git repository) can consume a great deal of memory; while not a leak per se this seems excessive and can cause failure if done on machines with limited ram. These can't be commonly occurring, or they'd be easier to nail down. On the other hand, I'd like to eliminate even rare memory leaks wherever possible. If any of you have concrete cases where you suspect excessive memory consumption (especially steadily increasing consumption, as with a leak) I'd be interested in any insight you can provide. Small test cases that use inordinate amounts of ram would be especially helpful. Thanks, -- Markus --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Puppet Developers" group. To post to this group, send email to puppet-dev@googlegroups.com To unsubscribe from this group, send email to puppet-dev+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/puppet-dev?hl=en -~----------~----~----~----~------~----~------~--~---
