Backport the commit
"2345407b8c29ab16d188eb15b78bff5f63632b7d MDEV-34716 Fix mysql.servers socket max length too short" which was pushed to 11.7. Also, see the test spider.alter_server in that commit for what it is fixing.

The reason for opening this issue rather than pushing it directly is because a direct backport causes a regression when running mtr --no-reorder rpl.rpl_mysql_upgrade,stmt rpl.rpl_mysqldump_gtid_slave_pos,stmt (see e.g. https://buildbot.mariadb.org/#/builders/369/builds/25168, at 0340a4a6aee bb-10.5-ycp-push):

worker[01] Using MTR_BUILD_THREAD 300, with reserved ports 19000..19029

rpl.rpl_mysql_upgrade 'stmt'             [ pass ]   2575

rpl.rpl_mysqldump_gtid_slave_pos 'stmt'  [ fail ]

        Test ended at 2024-10-15 13:18:56

CURRENT_TEST: rpl.rpl_mysqldump_gtid_slave_pos

--- /home/ycp/source/mariadb-server/10.5/src/mysql-test/suite/rpl/r/rpl_mysqldump_gtid_slave_pos.result	2024-10-14 17:39:33.249089152 +1100

+++ /home/ycp/source/mariadb-server/10.5/src/mysql-test/suite/rpl/r/rpl_mysqldump_gtid_slave_pos.reject	2024-10-15 13:18:55.785632591 +1100

@@ -36,12 +36,7 @@

 include/sync_with_master_gtid.inc

 select * from mysql.gtid_slave_pos;

 domain_id	sub_id	server_id	seq_no

-0	2	1	1

-0	3	1	2

-0	4	1	3

-0	5	1	4

-0	6	1	5

-0	7	1	6

+0	2	1	6

 select @@global.gtid_slave_pos as "after initial slave got in sync";

 after initial slave got in sync

 0-1-6

Result length mismatch

So it will require a bit more work. I can also confirm that this regression is not present in 11.7 that contains the patch.

BTW, regarding the magic number, see the glibc manual:

‘char sun_path[108]’
This is the file name to use.

Incomplete: Why is 108 a magic number? RMS suggests making
this a zero-length array and tweaking the following example to
use ‘alloca’ to allocate an appropriate amount of storage
based on the length of the filename.