Bug #76882 | table_io_waits_summary_by_index_usage does not show correct index names | ||
---|---|---|---|
Submitted: | 29 Apr 2015 11:24 | Modified: | 12 May 2015 16:30 |
Reporter: | Øystein Grøvlen | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Server: Performance Schema | Severity: | S3 (Non-critical) |
Version: | 5.7.6 | OS: | Any |
Assigned to: | CPU Architecture: | Any |
[29 Apr 2015 11:24]
Øystein Grøvlen
[29 Apr 2015 13:31]
Øystein Grøvlen
Posted by developer: Hi, I tried adding FLUSH TABLE t; FLUSH TABLE performance_schema.table_io_waits_summary_by_index_usage; FLUSH TABLES; to the script, but that did not change anything. I also tried the following: 1. Created table in 5.7.5 2. Ran query against table 3. Checked table_io_waits_summary_by_index_usage, it showed "idx" 4. Restarted server with 5.7.6 (NB! No upgrade script was run.) 5. Ran query against table 6. Checked table_io_waits_summary_by_index_usage, it showed "(index 0)"
[29 Apr 2015 14:14]
Øystein Grøvlen
Posted by developer: Got the script to work with FLUSH TABLES, but I needed to place the FLUSH after the execution of the query. If I restart the server and execute the query, I see "(index 0)" again. Another FLUSH TABLE makes it show the correct name. Does this mean that if index has not been access since server restart, a FLUSH TABLE is needed for existing connections to see the index name? That seems a bit awkward even in production.
[12 May 2015 16:30]
Paul DuBois
Noted in 5.7.8, 5.8.0 changelogs. The INDEX_NAME column of the Performance Schema table_io_waits_summary_by_index_usage table could sometimes show incorrect index names for tables until they had been in use for some time.