Description of Problem: After very little usage, (i.e. a couple of days) System can become unusable due to memory problems. (Problem with Riel VM?) We have two machines with RH 7.2 on them. Both of these have the latest update, but show symptoms with 2.4.9-7 also. (Never tested with 2.4.7-10) Our other machines with the exact same configuration, except using RH 7.1 with 2.4.3-12 kernel do not display this problem. These machines are PIII 800's with 512 RAM and 1024 swap. Typical usage includes emacs, galeon/mozilla, and gcc. My tool used to glean memory information is xosview. However, system slowdown and harddrive thrashing are evidence of the problem even if xosview is wrong. I may compile 2.4.15 when it comes out and see if the problem persists, and post an update. Version-Release number of selected component (if applicable): 2.4.9-13 How Reproducible: every time Steps to Reproduce: 1. boot up. 2. use computer for 1-3 days 3. notice slowdown and memory problems (heavy swap usage) Actual Results: computer slowed to barely usable state. swap memory nearly used up. (at 500M usage right now) Expected Results: swap memory should not be used up without being freed. I never ran programs that should have entered swap anyway, not with 512 MB of RAM anyway. Additional Information:
can you use "top" and then the "M" key to see if there's no accidental userspace memory leak ? (M sorts by memory usage)
Created attachment 37622 [details] a screenshot of top
I have attached a screenshot of top. It looks like it is a memory leak of some sort. Upon launch, galeon spawns 5 galeon-bin processes each taking 9% or 46M of memory. Additionally, I guess when I went to a certain page, 17 java_vm processes were launched, each taking 5% or 25M of memory. This constitutes 655M of memory usage! While this much memory use is appalling, when closing the applications down, the swap and memory usage still remains high, meaning there must be a memory leak of some sort. I know you don't distribute the jdk, and I am using galeon 12.7, so this is certainly not RH's responsibility, but, it looks like the mozilla, folks and galeon people need to work on this issue pretty badly. Thanks for your help, and any suggestions you might still have. But, I would have to say that although I didn't experience the grind down problems in 7.1, this probably isn't a kernel issue as much as memory issue with either mozilla's rendering engine used in galeon, or galeon itself coupled with how many java_vm's are launched each using 25M of memory. ouch! Thanks.