Open philpem opened 7 years ago
There are various database cleanups and cache operations which are done periodically, however when testing I've not had them use that much time or memory, even with quite large and old databases.
One thing that I did notice when testing though is that when loading an old database a rather excessive number of network connections are opened in parallel. Limiting the number of simultaneously open connections reduced the memory use quite significantly. I've pushed a commit 4883bfcd which does this.
I'll carry on looking and doing some more tests, to see if I can find anything else.
I found a memory corruption issue (fixed in dfc2e658) which may be related to this. Can you let me know if you still have the problem with this applied?
I will indeed! I'll leave my laptop off Twitter for a few days and reactivate the "really busy" account. That should give Retcon something to grind through.
Steps to reproduce:
This seems to be some kind of cache cleanup which is done in the background (as Retcon is fully functional up until the RAM runs out) and causes a ton of RAM to be consumed. Perhaps this cleanup could be done in stages to reduce the system load?
If this is what it is -- a progress indicator somewhere on the main screen might be nice too... "Cache cleanup, 0-100% done"