Ask Your Question
1

What is happening with high load average but mostly "wa"-wait?

asked 2017-01-10 01:04:06 -0600

wallyk gravatar image

Among several other annoyances since upgrading from F24 to F25, it is way too easy to get the system to stop responding.

The disk light is on solidly with only a wink (10 ms?) off every few seconds. I can't log in remotely (times out trying to connect) nor switch to an alternate console for diagnosis, but this time (after dozens of tries over the last 1.5 months) I finally got top to show this. Note the load average is climbing with one process running (presumably top) and 92% to 99% "waiting" (disk i/o wait?) cpu time shown each update for the duration.

top - 21:51:00 up 3 days,  6:28,  1 user,  load average: 6.62, 5.43, 2.84
Tasks: 200 total,   1 running, 198 sleeping,   0 stopped,   1 zombie
%Cpu0  :  0.3 us,  0.0 sy,  0.0 ni,  0.0 id, 99.3 wa,  0.0 hi,  0.3 si,  0.0 st
%Cpu1  :  1.3 us,  0.3 sy,  0.0 ni,  9.0 id, 89.4 wa,  0.0 hi,  0.0 si,  0.0 st
KiB Mem :  4037464 total,   230700 free,  3256972 used,   549792 buff/cache
KiB Swap:  4063228 total,  3690424 free,   372804 used.   219536 avail Mem 

  PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND     
 1588 wally     20   0 2235276 287484  32592 S   1.0  7.1 114:30.20 gnome-shell 
  327 root       0 -20       0      0      0 S   0.3  0.0   0:08.46 kworker/0:+ 
  899 boinc     30  10  350816   2328   1656 S   0.3  0.1   2:59.09 boinc_clie+ 
16370 wally     20   0 4234612 2.212g  41828 D   0.3 57.4 261:08.08 firefox     
27838 wally     20   0  155796   4084   3620 R   0.3  0.1   0:00.07 top         
    1 root      20   0  215140   3524   1056 S   0.0  0.1   0:06.12 systemd     
    2 root      20   0       0      0      0 S   0.0  0.0   0:00.13 kthreadd    
    3 root      20   0       0      0      0 S   0.0  0.0   0:01.23 ksoftirqd/0 
    5 root       0 -20       0      0      0 S   0.0  0.0   0:00.00 kworker/0:+ 
    7 root      20   0       0      0      0 S   0.0  0.0   0:45.58 rcu_sched   
    8 root      20   0       0      0      0 S   0.0  0.0   0:00.00 rcu_bh      
    9 root      20   0       0      0      0 S   0.0  0.0   0:15.13 rcuos/0     
   10 root      20   0       0      0      0 S   0.0  0.0   0:00.00 rcuob/0

I got the system to recover by closing Firefox which had been rendering two tabs which appeared blank, one containing this and the other this. After closing and waiting for the disk light to go off—taking 30-40 seconds—top shows

top - 21:58:00 up 3 days,  6:35,  1 user,  load average: 0.49, 2.56, 2.53
Tasks: 191 total,   1 running, 190 ...
(more)
edit retag flag offensive close merge delete

Comments

Try out iotop and smartctl if you suspect disk problems.

randomuser gravatar imagerandomuser ( 2017-01-10 01:26:33 -0600 )edit

Do remember having issue like that, removing the .mozilla folder in my home make firefox work fine, but you also have the same issue with google-chrome so not sure but you can give it a try.

aeperezt gravatar imageaeperezt ( 2017-01-10 07:57:14 -0600 )edit

2 Answers

Sort by » oldest newest most voted
1

answered 2017-01-23 15:49:30 -0600

wallyk gravatar image

I resolved the issue by reinstalling Fedora 24. On the same box, doing the same activities, it never exhibits the unresponsive behavior. F24 also resolves several other issues associated with F25 (no WiFi, USB to Android).

edit flag offensive delete link more
1

answered 2017-01-10 12:56:12 -0600

ssieb gravatar image

Processes in I/O wait contribute to the load average, that's why it's getting high like that. The Firefox process is causing a lot of swapping which will also cause other processes to end up in wait as well. I don't know why the sites would cause different browser behaviour in F25 vs. previous versions other than maybe different browser versions or possibly graphics drivers.

edit flag offensive delete link more

Comments

We can't rule out IO as a bottleneck, but this doesn't authoritatively point to swap thrashing IMO. Only ~400K of swap is in use. We need to see process IO statistics from iotop or iostat too.

randomuser gravatar imagerandomuser ( 2017-01-10 15:00:00 -0600 )edit
1

16370 wally 20 0 4234612 2.212g 41828 D 0.3 57.4 261:08.08 firefox and 219536 avail Mem tell me that it's definitely swapping. Also, that's 400MB of swap, not KB.

ssieb gravatar imagessieb ( 2017-01-10 16:38:57 -0600 )edit

facepalm - you're right, D state certainly isn't healthy. I wonder if it's firefox faulting, or if the physical disk is defective.

randomuser gravatar imagerandomuser ( 2017-01-12 23:23:03 -0600 )edit
1

D state is not unusual for Firefox. Depending on how many tabs you have open or what pages you have open, it can use a lot of memory. In this case I would not expect it to be a physical disk issue as the OP mentions everything works fine after closing Firefox.

ssieb gravatar imagessieb ( 2017-01-13 13:16:45 -0600 )edit

Question Tools

1 follower

Stats

Asked: 2017-01-10 01:04:06 -0600

Seen: 361 times

Last updated: Jan 23 '17