IPA high disk writes

[mhaluska]

Some idea why there is so high write load? Now it's 120GB written on VM serving only IPA server in 22 days.
There is almost no activity, 5 users registered, 1-2 active now.
Using iotop I found it's possible ns-slapd causing most of writes. Some idea how to decrease this?

Code: Select all

ipa-client-4.7.1-11.module_el8.0.0+79+bbd20d7b.x86_64
ipa-client-common-4.7.1-11.module_el8.0.0+79+bbd20d7b.noarch
ipa-common-4.7.1-11.module_el8.0.0+79+bbd20d7b.noarch
ipa-python-compat-4.7.1-11.module_el8.0.0+79+bbd20d7b.noarch
ipa-server-4.7.1-11.module_el8.0.0+79+bbd20d7b.x86_64
ipa-server-common-4.7.1-11.module_el8.0.0+79+bbd20d7b.noarch
ipa-server-dns-4.7.1-11.module_el8.0.0+79+bbd20d7b.noarch

I attached output from iotop -o -b -d 30

[hunter86_bg]

I can only recommend you to attach strace to your running process and check what it is writing to.

Code: Select all

strace -fvttTyyx -s 4096 -o /tmp/strace.txt -p <PID of process> 

Keep in mind that strace can slow things down , so you can't keep it running for ever.
If you want , you can try to filter only writes like this :

Code: Select all

strace -fvttTyyx -e trace=write -s 4096 -o /tmp/strace.txt -p <PID of process> 

[mhaluska]

Sorry for late response, I didn't get notification and then I totally forgot I created this post.
Output from iotop and strace are in attachments.

iotop.log.gz

(5.27 KiB) Downloaded 15 times

strace.txt.gz

(23.95 KiB) Downloaded 10 times