[YSQL] Major version upgrade failure on YBA universe

[foucher]

Jira Link: DB-13949

Description

Still investigating, but the main error from the upgrade logs is:

pg_restore: while PROCESSING TOC:
pg_restore: from TOC entry 2808; 1259 16402 INDEX XXXXidx yugabyte
pg_restore: error: could not execute query: ERROR:  could not serialize access due to concurrent update
Command was: CREATE INDEX "XXXXidx" ON "public"."XXXX" USING "lsm" ("id" HASH, "YYYY" ASC);

Seems possibly timing-related. The error occurs with 2-core nodes and not with 8-core nodes.

Could it be related to #24727?

Issue Type

kind/bug

Warning: Please confirm that this issue does not contain any sensitive information

• I confirm this issue does not contain any sensitive information.

[foucher]

From the YB master leader log, I believe this is the actual issue [line breaks/whitespace added]:

I1106 16:27:57.913406 58595 tablet_rpc.cc:495] Operation failed. Try again (yb/docdb/conflict_resolution.cc:87):
Failed Write(tablet: 1fe231164f3d489ebed535aec7d1c2ec, num_ops: 142, num_attempts: 1, txn: a5ae262a-d162-4959-8305-275b74d45282, subtxn: [none])
to tablet 1fe231164f3d489ebed535aec7d1c2ec
on tablet server { uuid: 40b73ecb495a4e71b9e6ded5f31b17e0
 private: [host: "10.9.218.183" port: 9100]
 cloud_info: placement_cloud: "aws" placement_region: "us-west-2" placement_zone: "us-west-2d"
after 1 attempt(s):
a5ae262a-d162-4959-8305-275b74d45282 conflicts with higher priority transaction: c8af857a-fa3e-4cdf-848a-6b0bccb06b15 (transaction error 3)

Since it says there are 142 ops it seems possible that it's the issue being fixed by Fizaa.

[foucher]

The issue didn't repro on b280 which contains 3cdb27c. Closing this.