Hi guys, I am troubleshooting an issue with `high...
# replication-ask-ai
d
Hi guys, I am troubleshooting an issue with
high CPU Usage of the EC2 instance hosting my Airbyte
but I cannot determine the root cause. ➡️ Context: • I used a
t2.large
to host only Airbyte. • I have several Airbyte jobs running every 15 minutes, several others on daily basic. • I observed that
cpu.user and cpu.stolen
were constantly at
45-50% each
, adding up to ~100% octavia shock (screenshot 1). ➡️ My quick solution: • I tried to restart all the running Airbyte containers (docker-compose down and up again). ➡️ After I restarted the Airbyte containers: • The
CPU.user
dropped to an avg. of
~ 5%
, compared to ~50% before. •
CPU.stolen
was
gone
. • For
testing
, I manually triggered some Airbyte jobs one by one and also together, these scenario caused some spikes,
up to 100% in case of CDC sync
in the CPU usage (screenshot 2 & 3) but nothing was unusual there octavia rolling eyes. • One step I missed was that I should have ran docker stats before restart them. Now I don't know the CPU usage of the Airbyte containers before the restart. ❓ Questions • Do you think this issue is specific to Airbyte? I read about why
cpu.stolen
could be high and it does not make sense in my case. The
cpu.stolen
is complete gone now (after the restart) while my Airbyte workload remains intact octavia thinking. • Should I implement some periodic restart for Airbyte? Thanks for your help. Daniel