728015 - Allocate physical Linux machine for manual Thunderbird update generation

Status: RESOLVED WONTFIX

(Release Engineering :: General, defect)

Description

[John Hopkins (:jhopkins)]

Manual partial update generation for Thunderbird 9.0.1->10.0.2 is taking much too long on our VM (momo-build-adm-01) and I'd like to allocate a physical Linux system for this purpose.
To give an idea of how bad the current system is, we're on update 162 of 204 after 6 hours.

Comment 1

[Rail Aliiev [:rail]]

* Updates will be generated manually, w/o slave-master connection
* Generated partial MARs will be uploaded to stage directly, using tb ssh keys and username
* Generated snippets will be transferred to aus2.mozillamessaging.com indirectly

Comment 2

[Rail Aliiev [:rail]]

John, how do you want to proceed with this? What about the following?

* let one of the regular build slaves start the updates builder
* wait until it start downloading complete MARs, so everything would be bumped and tagged
* stop the automation
* copy/paste clone/checkout commands on the fast slave (no bump/push needed)
* run patcher2.pl, make sure to run it in fast mode (--partial-patchlist-file=patchlist.cfg)
* follow regular steps

Comment 3

[Rail Aliiev [:rail]]

Per IRC: we can use a staging slave which can be easily allocated via slavealloc.