Why my large memory accessing program is not using Swap space ?
Issue
- On a 23.5 GB Memory system with 72 GB swap space the system does not make use of the swap space as it runs out of space. instead, the "offending" program requesting too much memory is killed.
- It turns out that when actually the available Memory value is exceeded swapping does not come into effect and instead the program trying to access swap space is killed.
- VM_overcommit parameters are also not working as per the kernel documentation.
- Provided memory hogger program is not able to acess the full swap space:
#OUTPUT OF MEMORY HOGGER PROGRAM#
10:09:47 got 22918 MiB
10:09:47 got 22919 MiB
10:09:47 got 22920 MiB
10:09:48 got 22921 MiB
10:09:48 got 22922 <-- Reached upto 23 GB around.
killed#OUTPUT OF SADC-COLLECTION SCRIPT#
#sar -f sadc-collection -tS | less10:08:52 AM kbswpfree kbswpused %swpused kbswpcad %swpcad
10:08:53 AM 73518708 2750852 3.61 1934976 70.34
10:08:54 AM 73518732 2750828 3.61 1934952 70.34
10:08:55 AM 73518732 2750828 3.61 1934952 70.34
10:08:56 AM 73518732 2750828 3.61 1934952 70.34
10:08:57 AM 73518732 2750828 3.61 1934952 70.34
10:08:58 AM 73518732 2750828 3.61 1934952 70.34
10:08:59 AM 73518732 2750828 3.61 1934952 70.34
10:09:00 AM 73518732 2750828 3.61 1934952 70.34
10:09:01 AM 73518732 2750828 3.61 1934952 70.34
10:09:02 AM 73518732 2750828 3.61 1934952 70.34
10:09:03 AM 73518732 2750828 3.61 1934952 70.34
10:09:04 AM 73518732 2750828 3.61 1934952 70.34
10:09:05 AM 73518732 2750828 3.61 1934952 70.34
10:09:06 AM 73518732 2750828 3.61 1934952 70.34
10:09:07 AM 73518732 2750828 3.61 1934952 70.34
10:09:08 AM 73555980 2713580 3.56 1897704 69.93
10:09:09 AM 75101536 1168024 1.53 319540 27.36
10:09:10 AM 74124364 2145196 2.81 1094992 51.04
10:09:12 AM 73117892 3151668 4.13 1923652 61.04
10:10:01 AM 73300560 2969000 3.89 1924316 64.81
10:10:08 AM 73300588 2968972 3.89 1924392 64.82
10:10:10 AM 73300604 2968956 3.89 1924452 64.82
10:10:14 AM 73305084 2964476 3.89 1939308 65.42
10:10:15 AM 73307916 2961644 3.88 1949096 65.81
10:10:16 AM 73309460 2960100 3.88 1960728 66.24
Hardly 4% of swap space has been used :(
- Changing the vm_overcommit parameters values does not make any diffrence and system fails to utilize the swap space.
- memory hogger program is eating memory so fast so is there a 'timing' issue in writing data too fast to swap partition causing system total inability to perform further paging at all ?
Environment
- Red Hat Enterprise Linux 6.2
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.