I have had a similar issue with logs in AG.

Believe it or not, I found that backing up the log three times in a row and then issuing a shrink works for me. You may also want to try backing up the log on the listener as well. I can assure it is possible to shrink the log in AG without removing the DB from the AG and re syncing.

I understand it's not best practice and I understand that it's odd...but I tried everything and it was the only thing that worked for me.

USE [pec_prod]

DBCC SHRINKFILE (N'mcr_dc_new_Log' , 0, TRUNCATEONLY)

GO

USE [pec_prod]

BACKUP LOG [pec_prod] TO DISK = N'nul' WITH NOFORMAT, INIT,

NAME = N'pec_prod_log Database Backup', SKIP, NOREWIND, NOUNLOAD, COMPRESSION, STATS = 10

GO

USE [pec_prod]

DBCC SHRINKFILE (N'mcr_dc_new_Log', 1528)

GO

USE [pec_prod]

BACKUP LOG [pec_prod] TO DISK = N'nul' WITH NOFORMAT, INIT,

NAME = N'pec_prod_log Database Backup', SKIP, NOREWIND, NOUNLOAD, COMPRESSION, STATS = 10

GO

USE [pec_prod]

DBCC SHRINKFILE (N'mcr_dc_new_Log' , 0, TRUNCATEONLY)

GO

BACKUP LOG [pec_prod] TO DISK = N'nul' WITH NOFORMAT, INIT,

NAME = N'pec_prod_log Database Backup', SKIP, NOREWIND, NOUNLOAD, COMPRESSION, STATS = 10

GO

Worth noting this was only on some of my instances. Some instances will backup the log and shrink no problem.