Extreme high load after opgrading to 8.17.1

[Normann P_ Nielsen _Netic_]

After upgrading to 8.17.1 the load on the ubuntu server spikes at +12 (normal around 2-5)

Automation took forever (50) min - https://confluence.atlassian.com/automation/upgrading-to-automation-for-jira-7-3-1055951956.html

We did not delete data first.

Also running Big Picture - opgrade tog +20 minutes for a "DB" upgrade.

Before and after:

[Normann P_ Nielsen _Netic_ · Accepted solution]

Stop Jira

Take backup

Execute on the Database:

TRUNCATE AO_589059_AUTOMATION_QUEUE;
TRUNCATE AO_589059_AUDIT_ITEM_PROJECT;
TRUNCATE AO_589059_AUDIT_ITEM_ASC_ITEM;
TRUNCATE AO_589059_AUDIT_ITEM_CGE_ITEM;
TRUNCATE AO_589059_AUDIT_ITEM_COMP_CGE;
TRUNCATE AO_589059_AUDIT_ITEM;
TRUNCATE AO_589059_RULE_STAT;
TRUNCATE AO_589059_RULE_STAT_ROLLUP_DAY;
TRUNCATE AO_589059_RULE_STAT_ROLLUP_HR;
TRUNCATE AO_589059_RULE_STAT_ROLLUP_MIN;

Start Jira.

Load stabilised on 1.X

And we have set Audit logm expiry to 4 weeks

[Normann P_ Nielsen _Netic_]

Disabling Jira Automation took load from +12 to 0-2 in 4-5 minutes

 

[Normann P_ Nielsen _Netic_]

This is problkably why:

jira=# select count(*) from "AO_589059_AUTOMATION_QUEUE";

 count 

-------

   129

(1 row)

jira=# select count(*) from "AO_589059_AUDIT_ITEM_PROJECT";

  count   

----------

 18307910

(1 row)

jira=# select count(*) from "AO_589059_AUDIT_ITEM_ASC_ITEM";

  count  

---------

 5571251

(1 row)

jira=# select count(*) from "AO_589059_AUDIT_ITEM_CGE_ITEM";

  count   

----------

 54987603

(1 row)

jira=# select count(*) from "AO_589059_AUDIT_ITEM_COMP_CGE";

  count   

----------

 57835510

(1 row)

jira=# select count(*) from "AO_589059_AUDIT_ITEM";

  count   

----------

 29693626

(1 row)

jira=# 

jira=# select count(*) from "AO_589059_RULE_STAT";

 count 

-------

 32077

(1 row)

jira=# select count(*) from "AO_589059_RULE_STAT_ROLLUP_DAY";

 count  

--------

 897587

(1 row)

jira=# select count(*) from "AO_589059_RULE_STAT_ROLLUP_HR";

 count  

--------

 216760

(1 row)

jira=# select count(*) from "AO_589059_RULE_STAT_ROLLUP_MIN";

 count 

-------

 32015

(1 row)

----

29 millions rows. Thats a lot. Unclear regarding these are index correctly

[Normann P_ Nielsen _Netic_]

Well, actually 54 millions in one table