Jira (PDB-2423) Limit initial sync to catalogs, factsets, and deactivate node commands

23 views
Skip to first unread message

Austin Blatt (JIRA)

unread,
Nov 4, 2019, 6:57:03 PM11/4/19
to puppe...@googlegroups.com
Austin Blatt updated an issue
 
PuppetDB / Improvement PDB-2423
Limit initial sync to catalogs, factsets, and deactivate node commands
Change By: Austin Blatt
Summary: Limit initial sync to latest catalogs , factsets, and reports deactivate node commands
Add Comment Add Comment
 
This message was sent by Atlassian JIRA (v7.7.1#77002-sha1:e75ca93)
Atlassian logo

Austin Blatt (JIRA)

unread,
Nov 4, 2019, 7:03:03 PM11/4/19
to puppe...@googlegroups.com

Austin Blatt (JIRA)

unread,
Nov 4, 2019, 7:03:03 PM11/4/19
to puppe...@googlegroups.com
Austin Blatt updated an issue
This should help replica provisioning To limit start up time while providing enough consistency for PuppetDB replicas, we need to limit the things amount of data that is transferred.

Currently
we care about sync
* Catalogs
* Factsets
* Reports
* Node deactivation
* Catalog Inputs (for cd4pe)

By far the two largest sets of data are reports and catalog inputs, so I recommend that initial sync is limited to the following.
* Catalogs
* Factsets
* Node deactivation

The most commonly suggested modification to this list is to sync _only_ the latest reports. I haven't heard a compelling reason to spend startup time syncing the latest report for a replica PuppetDB because all the time PDB spends in startup sync it is drifting out of alignment. If someone has a good reason that a replica PuppetDB should have the latest reports when it starts up, I would be happy to add that to the list of things to sync.

Austin Blatt (JIRA)

unread,
Nov 4, 2019, 7:04:03 PM11/4/19
to puppe...@googlegroups.com
Austin Blatt updated an issue
To limit start up time for PuppetDB replicas, we need to limit the amount of data that is transferred.

Currently we sync

* Catalogs
* Factsets
* Reports
* Node deactivation
* Catalog Inputs (for cd4pe)

By far the two largest sets of data are reports and catalog inputs, so I recommend think that initial sync is should be limited to the following to ensure a somewhat fast startup time .

* Catalogs
* Factsets
* Node deactivation

The most commonly suggested modification to this list is to also sync _only_ the latest reports. I haven't heard a compelling reason to spend startup time syncing the latest report for a replica PuppetDB because all the time PDB spends in startup sync it is drifting out of alignment. If someone has a good reason that a replica PuppetDB should have the latest reports when it starts up, I would be happy to add that to the list of things to sync.

Charlie Sharpsteen (JIRA)

unread,
Dec 5, 2019, 6:47:05 PM12/5/19
to puppe...@googlegroups.com

Charlie Sharpsteen (JIRA)

unread,
Dec 5, 2019, 6:52:05 PM12/5/19
to puppe...@googlegroups.com

Jarret Lavallee (JIRA)

unread,
Dec 5, 2019, 7:35:06 PM12/5/19
to puppe...@googlegroups.com

Nick Burgan (JIRA)

unread,
Jan 15, 2020, 2:27:04 PM1/15/20
to puppe...@googlegroups.com

Nick Burgan (JIRA)

unread,
Jan 15, 2020, 2:28:04 PM1/15/20
to puppe...@googlegroups.com

Nick Burgan (JIRA)

unread,
Jan 15, 2020, 2:28:04 PM1/15/20
to puppe...@googlegroups.com

Nick Burgan (JIRA)

unread,
Jan 29, 2020, 2:04:05 PM1/29/20
to puppe...@googlegroups.com
Nick Burgan updated an issue
Change By: Nick Burgan
Sprint: HA Team 2020-01-29 , HA Team 2020-02-12

Austin Blatt (JIRA)

unread,
Feb 5, 2020, 5:55:04 PM2/5/20
to puppe...@googlegroups.com
Austin Blatt assigned an issue to Unassigned
Change By: Austin Blatt
Assignee: Austin Blatt

Melissa Amos (JIRA)

unread,
Feb 7, 2020, 1:42:05 PM2/7/20
to puppe...@googlegroups.com

Nick Burgan (JIRA)

unread,
Feb 12, 2020, 2:05:06 PM2/12/20
to puppe...@googlegroups.com
Nick Burgan updated an issue
Change By: Nick Burgan
Sprint: HA Team 2020-01-29, HA Team 2020-02-12 , HA Team 2020-02-26

Austin Blatt (JIRA)

unread,
Feb 20, 2020, 6:08:17 PM2/20/20
to puppe...@googlegroups.com
Reply all
Reply to author
Forward
0 new messages