Query: Ezmeral installation error

[SaveethaR]

While installing Ezmeral, after configuring, cluster and nodes, while verification it fails with the below error , but the host is reachable

getent ahosts vfn-spdr-lab1-ezm02
xx.xxx.xxx.xx STREAM vfn-spdr-lab1-ezm02
xx.xxx.xxx.xx DGRAM
xx.xxx.xxx.xx RAW

2024-10-21 14:48:59.045: IGNORED Task: Measure DNS resolution time(shell {
    "_raw_params": "start=$(date +%s%N)\ngetent ahosts {{ item }}\nend=$(date +%s%N)\necho \"$((($end - $start)/1000000)) ms\"\n"
}) -> failed  {
    "changed": false
}
One or more items failed in Measure DNS resolution time(shell {
    "_raw_params": "start=$(date +%s%N)\ngetent ahosts {{ item }}\nend=$(date +%s%N)\necho \"$((($end - $start)/1000000)) ms\"\n"
})
vfn-spdr-lab1-ezm03: ...ignoring errors
2024-10-21 14:48:59.092: Task: Check if resolution time exceeds 300ms(fail {
    "msg": "DNS resolution for {{ item.item }} took more than 300ms ({{ item.stdout_lines[-1] }})"
}) -> failed  =>  {
    "_ansible_no_log": false
}
The conditional check 'item.stdout_lines[-1] | regex_replace(' ms', '') | int > 300' failed. The error was: error while evaluating conditional (item.stdout_lines[-1] | regex_replace(' ms', '') | int > 300): 'dict object' has no attribute 'stdout_lines'

The error appears to be in '/opt/mapr/installer/mapr_ansible/playbooks/check.yml': line 144, column 9, but may
be elsewhere in the file depending on the exact syntax problem.

The offending line appears to be:


      - name: Check if resolution time exceeds 300ms
        ^ here

[Dave Olker]

That appears to be a transient issue with your DNS server.  If you retry the installation does it still fail with DNS resolution taking too long?  I've seen this in my environment and usually retrying the install works.  If not, you can try configuring your hosts with local entries in the /etc/hosts file and configuring your name service to resolve hosts using /etc/hosts instead of DNS, at least until the installation succeeds.

[Shishir_Prakash]

It's probably not the product issue.   Mostly name resolution issue. Do you use proxy ? If yes , this could be proxy config issue.  Here is what I will suggest to get rid of DNS issue. 

- On all nodes trying manually adding IP/host of all the cluster nodes under /etc/hosts

- Validate if the proxy is configured correctly to be used by installer.  Check below doc for proxy section. 

https://docs.ezmeral.hpe.com/datafabric-customer-managed/78/AdvancedInstallation/c_install_prerequisites.html

[SaveethaR]

@Shishir_Prakash  @Dave Olker 

yes tried adding the hosts in /etc/hosts file in all 3 nodes. still it is failing. But the below block is failing, but the command getent ahosts  is getting executed successfully.

- name: Check DNS resolution time for each host in the cluster
block:
- name: Measure DNS resolution time
shell: |
start=$(date +%s%N)
getent ahosts {{ item }}
end=$(date +%s%N)
echo "$((($end - $start)/1000000)) ms"
register: dns_resolution_time
loop: "{{ groups['all'] }}"
ignore_errors: yes

- name: Check if resolution time exceeds 300ms
fail:
msg: "DNS resolution for {{ item.item }} took more than 300ms ({{ item.stdout_lines[-1] }})"
loop: "{{ dns_resolution_time.results }}"
when: item.stdout_lines[-1] | regex_replace(' ms', '') | int > 300

- name: Display DNS resolution times
debug:
msg: "DNS resolution time for {{ item.item }} was {{ item.stdout }}"
loop: "{{ dns_resolution_time.results }}"

[Shishir_Prakash]

I will suggest you to validate your proxy settings and if possible on all nodes add below line in /etc/sudoers . Is there is any permission issue, it will take care. 

mapr ALL=(ALL) NOPASSWD: ALL

[SaveethaR]

Hi,

DNS error is resolved. Thank you. As we are doing installation in Development environment, the pre-requisite for /tmp (10GB) and /opt/mapr(128GB) is failing. Can we reduce the pre-requisite space for /tmp and /opt/mapr. If yes, can you let us know, where this has to be updated. Thank you.

[Mahesh_S]

As per the prerequisites, provide at least 10 GB of free disk space on the operating system partition,  10 GB of free disk space in the /tmp  directory and 128 GB of free disk space in the /opt  directory. Services such as the ResourceManager and NodeManager use the /tmp  directory. We cannot customize these prerequisite values.

[Shishir_Prakash]

You can try setting java.io.tmpdir to a different directory with sufficient space and see if that resolves the issue. Refer to the document below for details:

https://docs.ezmeral.hpe.com/datafabric-customer-managed/78/AdvancedInstallation/c_install_prerequisites.html

Note: I recommend marking the comment that helped resolve the issue. This improves the algorithm's ability to suggest accurate solutions and makes it easier for others to find.

[SaveethaR]

Thank you for the response. Is /opt/mapr also requires 128GB? as the installation is erroring out due this.

[Dave Olker]

Detailed node requirements are here: https://docs.ezmeral.hpe.com/datafabric-customer-managed/78/AdvancedInstallation/PreparingEachNode.html

Detailed information specific to RAM and disk space are here: https://docs.ezmeral.hpe.com/datafabric-customer-managed/78/AdvancedInstallation/PreparingEachNode-memory.html

[support_s]

Hello,

Let us know if you were able to resolve the issue.

If you have no further query, and you are satisfied with the answer then kindly mark the topic as Solved so that it is helpful for all community members.

Please click on "Thumbs Up/Kudo" icon to give a "Kudo".

Thank you for being a HPE valuable community member.