Out of memory again

64GB DDR4 RAM. I have not yet upgraded to a new system with 128GB DDR5 RAM but I am about to. I may even reduce my storage amount to below the RAM amount to make running out an impossibility. Then I will finallly have a decent system and not be such a bother around here. Jason_25 runs out of RAM again. How boring.

I now have my system in a state where I can only open around 1 tab at a time in the browser. Also killall -9 falkon is unsuccesful most of the time. This command should not return to the cursor until complete but it does anyway and falkon is not killed. Since I have the system in this state and it has not crashed yet or rebooted what can I do to figure out what is going on? I assume the ZFS arc cache should still be allowing me to open at least 1 tab with this much memory?

"top -o res -d 1":
Code:
last pid: 22643;  load averages:  1.43,  1.17,  0.98   up 104+09:26:55 10:47:46
106 processes: 1 running, 105 sleeping
CPU:  5.4% user,  0.6% nice,  1.7% system,  0.3% interrupt, 92.0% idle
Mem: 37M Active, 3216K Inact, 2464K Laundry, 57G Wired, 480M Free
ARC: 1907M Total, 1026M MFU, 188M MRU, 266K Anon, 18M Header, 673M Other
     804M Compressed, 2921M Uncompressed, 3.63:1 Ratio

  PID USERNAME    THR PRI NICE   SIZE    RES STATE    C   TIME    WCPU COMMAND
11404 jason        29  21    0  1484M   699M uwait    6 150:08  16.99% mpv
26032 jason        30  21    0  1702M   597M uwait    5 430.5H  16.16% mpv
22541 jason        70  21    0  1133M   344M select   2   0:48   3.47% falkon
25807 jason        14  26    0   535M   242M select   4  78.3H   6.64% Xorg
21976 jason        73  20    0  3048M   235M select   4   0:21   0.59% firefox
25861 jason        50  21    0  1141M   152M select   7  14:30   4.00% plasmash
22575 jason        19  36   10    16G    82M uwait    2   0:13   0.78% QtWebEng
22560 jason        12  36   10  8657M    81M uwait    0   0:02   0.00% QtWebEng
25841 jason        18  20    0   651M    78M select   1  15:14   0.49% kwin_x11
21982 jason        29  20    0  2580M    73M select   7   0:04   0.00% firefox
 2316 jason        32  20    0   656M    72M select   6   0:14   0.00% krunner
22561 jason        12  36   10  8659M    60M uwait    5   0:05   1.27% QtWebEng
25977 jason        16  20    0   542M    57M select   3   1:05   0.00% konsole
22638 jason        13  68    0   472M    49M select   5   0:01  59.67% konsole
21985 jason        29  20    0  2534M    38M select   7   0:03   0.59% firefox
21983 jason        28  20    0  2512M    38M select   0   0:01   0.00% firefox
25956 jason         9  20    0   457M    37M select   2 284:31   0.49% konsole
34925 jason        10  20    0   479M    36M select   0   0:22   0.00% konsole
26435 jason        10  20    0   457M    35M select   2  48:52   0.00% konsole
25995 jason         9  20    0   461M    35M select   4 412:12   0.00% konsole
26015 jason         9  20    0   467M    35M select   2 351:41   0.00% konsole
26310 jason         9  20    0   456M    34M select   5 102:43   0.00% konsole
26253 jason         9  20    0   457M    33M select   4   1:37   0.00% konsole
57739 jason        10  20    0   452M    30M select   2 170:44   0.00% ksysguar
25869 jason        10  20    0   427M    24M select   5  11:57   0.00% Discover
25855 jason         9  20    0   327M    23M select   1   0:44   0.00% ksmserve
25829 jason         9  20    0   318M    22M select   0   0:28   0.00% klaunche
25860 jason         9  20    0   320M    21M select   2   0:40   0.00% kaccess
21988 jason        19  20    0  2446M    21M select   3   0:00   0.00% firefox
21987 jason        19  20    0  2446M    21M select   5   0:00   0.00% firefox
 
Run vmstat -m on a freshly booted system. Save the output somewhere. Run it again later and again later. Compare the outputs. If there's a kernel memory leak it will help you discover it.

Probably a good idea to run the above diagnostics and open a bugzilla bug about this.

Also describe what kind of work you do on this system. If it is a bug, it will help track it down. If it's not a bug, it will explain what the workload is causing the kernel to allocate memory as it does.

But without the diagnostics all anyone can do is guess. And guessing doesn't solve problems.
 
Back
Top