Let undercloud-process-count succeed with warnings
Some services set the number of their processes equal to the number of CPUs, which poses the risk of running out of memory in cases where this number is very high. This validation is intended to catch these cases and therefore breaks if the number of processes is higher than 8. However, the validation in its current form cannot make a judgement for a realistic limit for any given situation. This patch changes the validation so it doesn't fail when the limit is reached but succeeds with warnings. For a follow-up change we should investigate if increasing the limit might be reasonable, or even if there might be some heuristics to determine the limit dynamically based on the undercloud machine's characteristics. Change-Id: I68e25e2a2f7f840144de7cab97731dde3aeb3694
This commit is contained in:
parent
65a52d662a
commit
02000cccff
|
@ -32,7 +32,14 @@
|
|||
- swift-container-server
|
||||
- zaqar-server
|
||||
|
||||
- name: Verify the number of running processes per OpenStack service
|
||||
fail: msg="There are {{ item.stdout }} {{ item.item }} processes running. Having more than {{ max_process_count }} risks running out of memory."
|
||||
failed_when: "{{ item.stdout|int }} > {{ max_process_count|int }}"
|
||||
- name: Create warning messages
|
||||
command: echo "There are {{ item.stdout }} {{ item.item }} processes running. Having more than {{ max_process_count }} risks running out of memory."
|
||||
register: process_warnings
|
||||
with_items: "{{ process_count.results }}"
|
||||
when: "{{ item.stdout|int }} > {{ max_process_count }}"
|
||||
|
||||
- name: Output warning message
|
||||
warn: msg={{ warning_msg }}
|
||||
when: "{{ warning_msg|length > 0}}"
|
||||
vars:
|
||||
warning_msg: "{{ process_warnings.results|selectattr('changed')|map(attribute='stdout')|join('\n') }}"
|
||||
|
|
Loading…
Reference in New Issue