https://www.puppet.com/community logo
Join Slack
Powered by
# puppet-enterprise
  • b

    bastelfreak

    09/15/2022, 7:13 PM
    curl -X POST <http://127.0.0.1:8080/pdb/admin/v1/cmd|http://127.0.0.1:8080/pdb/admin/v1/cmd> -H 'Accept: application/json' -H 'Content-Type: application/json' -d '{"command": "clean", "version": 1, "payload": ["other"] }'
  • b

    bastelfreak

    09/15/2022, 7:13 PM
    neat
  • r

    ramindk

    09/15/2022, 7:24 PM
    @Reid Conti ^^^^ feels similar to issues we struggled with last year. I want to say we were running cleanups more often, but I forget the details. Hourly cleanup sound familiar and different from expiring daily reports.
  • b

    bastelfreak

    09/15/2022, 7:28 PM
    we have a gc-interval of 60 minutes but the
    other
    task runs once per day, at most. very odd
  • s

    Slackbot

    09/15/2022, 7:29 PM
    This message was deleted.
    r
    s
    • 3
    • 3
  • b

    bastelfreak

    09/16/2022, 3:31 PM
    update on that: we created a temporary table with all resources that need to go:
    Copy code
    EXPLAIN ANALYZE CREATE TABLE bad_kitties AS
    SELECT
        resource
    FROM
        resource_params_cache
    EXCEPT
    SELECT
        resource
    FROM
        catalog_resources;
    
    CREATE INDEX bad_kitties_idx ON bad_kitties (resource);
    And we're deleting them 250k batches:
    Copy code
    EXPLAIN ANALYZE
    WITH loosers AS (
        SELECT
            *
        FROM
            bad_kitties
        ORDER BY
            resource
        LIMIT 250000
    ), dead_loosers AS (
        DELETE FROM bad_kitties AS bk
        USING
            loosers AS l
        WHERE
            bk.resource = l.resource
    )
    DELETE FROM resource_params_cache AS rpc
    USING
        loosers AS l
    WHERE
        rpc.resource = l.resource;
    I'm open for suggestions
  • b

    bastelfreak

    09/16/2022, 3:31 PM
    update on that: we created a temporary table with all resources that need to go:
    Copy code
    EXPLAIN ANALYZE CREATE TABLE bad_kitties AS
    SELECT
        resource
    FROM
        resource_params_cache
    EXCEPT
    SELECT
        resource
    FROM
        catalog_resources;
    
    CREATE INDEX bad_kitties_idx ON bad_kitties (resource);
    And we're deleting them 250k batches:
    Copy code
    EXPLAIN ANALYZE
    WITH loosers AS (
        SELECT
            *
        FROM
            bad_kitties
        ORDER BY
            resource
        LIMIT 250000
    ), dead_loosers AS (
        DELETE FROM bad_kitties AS bk
        USING
            loosers AS l
        WHERE
            bk.resource = l.resource
    )
    DELETE FROM resource_params_cache AS rpc
    USING
        loosers AS l
    WHERE
        rpc.resource = l.resource;
    I'm open for suggestions
  • s

    Slackbot

    09/19/2022, 11:21 PM
    This message was deleted.
    d
    m
    • 3
    • 4
  • b

    bastelfreak

    09/21/2022, 5:33 PM
    I can take a look when I'm back in the office in two weeks. would that result in smaller HEAP?
  • r

    Robert Vincent

    09/29/2022, 1:18 PM
    After completely wiping and reinstalling CD4PE control node, we discovered that a bug in the latest version makes it impossible to setup jobs that run outside of Docker. I received an email today saying "Puppet Application Manager 1.81.1 is now available!". It referenced this link for installation and upgrade procedures: https://puppet.com/docs/continuous-delivery/4.x/cd-pam-index.html From there, I followed the "Upgrading PAM on a Puppet-supported cluster" link: https://puppet.com/docs/continuous-delivery/4.x/pam/upgrade-pam.html#upgrade-pam That page instructed me to run the following command:
    kubectl get installer --sort-by=.metadata.creationTimestamp -o jsonpath='{.items[-1:].metadata.name}' ; echo
    The command output
    merged
    , which according to the instructions means that we are running "Legacy architecture". Does this seem correct to you? Why would a newly-installed PAM be classified as "Legacy architecture"?
  • c

    CVQuesty

    09/29/2022, 1:41 PM
    Why would one upgrade PAM when it’s tied to the core auth framework in Linux?
  • r

    Robert Vincent

    09/29/2022, 1:41 PM
    In this context, PAM = Puppet Application Manager, not Pluggable Authentication Modules.
  • c

    CVQuesty

    09/29/2022, 1:42 PM
    oh… WHEW!!!
  • c

    CVQuesty

    09/29/2022, 1:42 PM
    YOu see how much CD4PE I’m into… 😄
  • h

    hbui

    09/29/2022, 2:14 PM
    coming in blind, I'd wonder whether it was saying that it was a Legacy kubernetes set up
  • r

    Robert Vincent

    09/29/2022, 2:27 PM
    According to the docs, "Legacy" means "Installed prior to May 21, 2021". We installed in September, 2022.
  • h

    hbui

    09/29/2022, 2:28 PM
    date of install is a really odd metric
  • r

    Robert Vincent

    09/29/2022, 2:28 PM
    (shrug)
  • r

    Robert Vincent

    09/29/2022, 2:28 PM
    I didn't write the docs.
  • h

    hbui

    09/29/2022, 2:30 PM
    which version of kubernetes are you running?
  • h

    hbui

    09/29/2022, 2:32 PM
    the upgrade doc you link suggests to me that legacy means using Rook 1.0
  • h

    hbui

    09/29/2022, 2:34 PM
    but yes, very odd that grabbing a new version says that you are using "Legacy"
  • r

    Robert Vincent

    09/29/2022, 2:34 PM
    According to the installation logs. I was running version 1.22 prior to attempting the upgrade.
  • r

    Robert Vincent

    09/29/2022, 2:35 PM
    BTW, the upgrade failed. Restoring from snapshot also failed. Looks like we need to do yet another complete wipe and reinstall.
  • h

    hbui

    09/29/2022, 2:35 PM
    I'd probably be inclined to get support services on the line
  • c

    CVQuesty

    09/29/2022, 2:39 PM
    this
  • r

    Robert Vincent

    09/29/2022, 2:48 PM
    Yeah, I've got two tickets open.
  • r

    Robert Vincent

    09/29/2022, 2:52 PM
    I expect that after two or three weeks of back-and-forth suggestions, I'm either going to ditch CD4PE altogether or do another complete wipe and reinstall onto a freshly-provisioned VM. Again.
  • d

    David Sandilands

    09/29/2022, 3:58 PM
    Hey @Robert Vincent I appreciate this is frustrating, I'm an architect for Pupept and on one of our internal Puppet channels where support, architecture and other teams come together and review cases to get a better understanding of issues coming up on the CD4PE platform and how to better document and approach CD4PE implementation to avoid/fix them. I've pinged details into our channel and I will see what we can find.
    👍 3
  • r

    Robert Vincent

    09/29/2022, 4:00 PM
    I expect that after two or three weeks of back-and-forth suggestions, I'm either going to ditch CD4PE altogether or do another complete wipe and reinstall onto a freshly-provisioned VM. Again.
1...262728...73Latest