sessions/challenges-after-upgrading-to-3.5-from-2.7.md - summit/2024 - Git at Google

 # Challenges after upgrade to 3.5 from 2.7

 Last year, Qualcomm migrated its largest Gerrit instance from 2.7 to
 3.5. In a previous talk, we discussed our optimizations to schema
 upgrades, NoteDB migration, and offline reindexing. In this talk,
 I’ll delve into the challenges we faced close to the upgrade date and
 after going live. I’ll focus mainly on the issues we encountered with
 our multi-primary setup, repositories on NFS, and JGit, as well as our
 solutions. Also, I’ll explore the performance problems we faced and how
 we overcame them.

 I’ll also share some positives, such as what worked well after the
 upgrade, including our use of Elasticsearch as the index backend
 and some updates we made to the Elasticsearch module to meet our needs.

 *[Kaushik Lingarkar, Engineer / Qualcomm Technologies, Inc](../speakers.md#kaushikl)*
	# Challenges after upgrade to 3.5 from 2.7

	Last year, Qualcomm migrated its largest Gerrit instance from 2.7 to
	3.5. In a previous talk, we discussed our optimizations to schema
	upgrades, NoteDB migration, and offline reindexing. In this talk,
	I’ll delve into the challenges we faced close to the upgrade date and
	after going live. I’ll focus mainly on the issues we encountered with
	our multi-primary setup, repositories on NFS, and JGit, as well as our
	solutions. Also, I’ll explore the performance problems we faced and how
	we overcame them.

	I’ll also share some positives, such as what worked well after the
	upgrade, including our use of Elasticsearch as the index backend
	and some updates we made to the Elasticsearch module to meet our needs.

	[Kaushik Lingarkar, Engineer / Qualcomm Technologies, Inc](../speakers.md#kaushikl)