0

I have a node alpine docker container (hosted on Azure) which periodically reaches 100% CPU usage. However, running the top or htop command shows no threads having high CPU usage.

How do I find out what the cause is? It seams like the node process isn't causing the high CPU usage. I also ran the node profiler and the cpu profile from the times when CPU is 100% has no obvious change to a regular profile.

htop screen shows that CPU is at 100% (mostly blue, meaning low priority threads). But the threadlist below only shows 17% usage for node.

The high CPU usage is quite periodic: CPU usage over time. And it started a couple days ago after an deployment, but there were no significant code changes (and reverting did not fix the issue).

Edit:

vmstat (first during normal cpu usage, 2nd during 100% CPU)

ae7d7f5ef9c8:~# vmstat 1 5
procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu-----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa st
 7  0 457816 296204 275828 1383336    4    6   167   571    0   17 33 13 54  1  0
 0  0 457816 296576 275860 1383388    0    0     0   256    0 4297 30 14 56  0  0
 0  0 457816 296576 275908 1383340    0    0     0   296    0 4301 23  8 69  0  0
 0  0 457816 296576 275932 1383348    0    0     0   208    0 2985 22  8 70  0  0
 0  0 457816 296576 275940 1383348    0    0     0    80    0 2444 21  4 75  0  0

ae7d7f5ef9c8:~# vmstat 1 5
procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu-----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa st
 3  0 459432 285672 259408 1331844    4    6   167   571    0   23 33 13 54  1  0
 1  0 459432 247584 259420 1331860    0    0     0   124    0 4078 76 24  0  0  0
 1  0 459432 122716 259468 1331988    0    0     0   300    0 5521 64 36  0  0  0
 2  0 459432 282488 259504 1332020    0    0     0   212    0 5351 76 24  0  0  0
 5  0 459432 228736 259596 1332024    0    0     0   340    0 3257 76 24  0  0  0
Improve this question
4
  • what is the result of vmstat 1 5?
    – Romeo Ninov
    Feb 8 at 14:36
  • system resource bottlenecks such as disk I/O, network I/O may cause CPU peak
    – Zareh Kasparian
    Feb 8 at 19:33
  • The application has no disk I/O operation except logging and htop doesn't show significant usage in regards to disk IO.
    – zapdev
    Feb 9 at 8:13
  • Regarding network I/O: The issue is also during night with <1000 request/minute. (During day 10k requests/minute)
    – zapdev
    Feb 9 at 8:15

1 Answer 1

Reset to default
0

Update: The issue seem to be with Azure Webapp platforms. After creating a new Webapp an deploying there, the issue is gone.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .