API logs not generated in 4.20

[Pearl1594]

problem

Most likely due to migration to log4j2, it appears that api logs aren't getting generated in ACS 4.20.0

versions

4.20

The steps to reproduce the bug

What to do about it?

No response

[gpordeus]

Hi @Pearl1594 !

I checked it out and the API logger ended up being combined with the general logger by mistake. I'll open the fix right now.

[DaanHoogland]

@gpordeus just doing now

[andrijapanicsb]

I had a case when upgrading from 4.19.x to 4.20, on Debian, when answering "no" when asked to overwrite the log4j (and /etc/default/cloudstack-management file) - the upgraded environment was unable to start the management server).

Not sure how we handled this previously on previous Ubuntu versions - was there still an option for the admin to "keep" the current files (when asked if we should overwrite them or keep the existing ones)

[DaanHoogland]

@andrijapanicsb , we never had an incompatible log4j configuration before.

[JoaoJandre]

I had a case when upgrading from 4.19.x to 4.20, on Debian, when answering "no" when asked to overwrite the log4j (and /etc/default/cloudstack-management file) - the upgraded environment was unable to start the management server).

Not sure how we handled this previously on previous Ubuntu versions - was there still an option for the admin to "keep" the current files (when asked if we should overwrite them or keep the existing ones)

@andrijapanicsb If you decide to not upgrade automatically, you're responsible for upgrading it manually before starting the management server, as log4j 1.29 and log4j 2 configuration descriptions are incompatible.

[andrijapanicsb]

@JoaoJandre thx, that is correct, but I have a bigger problem of "/etc/default/cloudstack-management" not being replaced by default (at least when using Ansible) and that would cause 4.20 mgmt server to not boot at all (not even come to the point of trying to establish a DB connection). This is a puzzle for me on "why we allow" this and if there is anything that can be done (and I understand this is now an off-topic to this original issue). Thx

[weizhouapache]

@JoaoJandre thx, that is correct, but I have a bigger problem of "/etc/default/cloudstack-management" not being replaced by default (at least when using Ansible) and that would cause 4.20 mgmt server to not boot at all (not even come to the point of trying to establish a DB connection). This is a puzzle for me on "why we allow" this and if there is anything that can be done (and I understand this is now an off-topic to this original issue). Thx

"/etc/default/cloudstack-management" was updated in #8497

some java opts are required when run ACS with java 17, users have two ways to bypass it

• use java 11 instead of java 17. I do not recommend it
• update "/etc/default/cloudstack-management" manually, refer to https://github.com/apache/cloudstack/blob/main/packaging/systemd/cloudstack-management.default

the /etc/default/cloudstack-usage needs to be updated as well, if it is not overwritten.