gtids executed are missing from mysql.gtid_executed table

[lin yuan]

Hello experts

I am running a master - slave replication based on GTID in my production environment.
- mysql-community-server-5.7.17
- centos 7.4 x86_64

Today, I notice the Executed_Gtid_Set from "show slave status\G" is not consistent with info from mysql.gtid_executed

the output excerpt from "show slave status\G":
Retrieved_Gtid_Set: baddb4d3-6c52-11e7-bb60-00505690d333:11-6540560
Executed_Gtid_Set: baddb4d3-6c52-11e7-bb60-00505690d333:1-6540560

the output from "select * from mysql.gtid_executed;":
+--------------------------------------+----------------+--------------+
| source_uuid | interval_start | interval_end |
+--------------------------------------+----------------+--------------+
baddb4d3-6c52-11e7-bb60-00505690d333 | 11 | 5885134

I know the values from gtid_executed table can be different from "show slave status\G", but it does not look good the interval_start is "11" (1-10 are missing).
When I tried to restart mysqld, the replication failed to start with error 1236 in log:

2017-10-20T03:36:48.559146Z 1 [ERROR] Slave I/O for channel '': Got fatal error 1236 from master when reading data from binary log: 'The slave is connecting using CHANGE MASTER TO MASTER_AUTO_POSITION = 1, but the master has purged binary logs containing GTIDs that the slave requires.', Error_code: 1236

And then I checked slave status. The "show slave status\G" shows
Retrieved_Gtid_Set: baddb4d3-6c52-11e7-bb60-00505690d333:11-6540733
Executed_Gtid_Set: baddb4d3-6c52-11e7-bb60-00505690d333:11-6540733

It looks the gtid_executed global variable is initialized from mysql.gtid_executed table after a restart? and then the slave requests gtids baddb4d3-6c52-11e7-bb60-00505690d333:1-10 from master, which were purged long time ago.

So do you have any idea why gtids baddb4d3-6c52-11e7-bb60-00505690d333:1-10 are missing from mysql.gtid_executed table? I remembered the record with interval_start of 1 existed some days ago.