Provide fixes for MySQL bugs #51196 / #61790 in Percona Server

Bug #813587 reported by Alexey Kopytov
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
MySQL Server
Unknown
Unknown
Percona Server moved to https://jira.percona.com/projects/PS
Fix Released
Undecided
Alexey Kopytov
5.1
Fix Released
Undecided
Alexey Kopytov
5.5
Fix Released
Undecided
Alexey Kopytov

Bug Description

We need to provide fixes for http://bugs.mysql.com/bug.php?id=51196 and http://bugs.mysql.com/bug.php?id=61790 in Percona Server 5.1 and 5.5.

Tags: cr i12461

Related branches

Revision history for this message
Alexey Kopytov (akopytov) wrote :

Trying to link the second bug via remote bug watch: http://bugs.mysql.com/bug.php?id=61790

Revision history for this message
Alexey Kopytov (akopytov) wrote :

Cannot target to multiple series with multiple affected projects.

Changed in percona-server:
assignee: nobody → Alexey Kopytov (akopytov)
milestone: none → 5.1.58-12.9
Revision history for this message
Alexey Kopytov (akopytov) wrote :

Duplicating my analysis from http://bugs.mysql.com/bug.php?id=61790 here:

- when putting a table (not necessarily a FEDERATED one) to the table cache, server may
evict and close another if the table cache is full
- if the victim happens to be a FEDERATED table, its client connection to remote server
is closed
- if the remote server has gone for whatever reasons (stopped, connection timed out,
network issues), mysql_close() in ha_federated::close() fails with ER_NET_ERROR_ON_WRITE
- the above error is propagated back to the client that issued the original query, even
if it was against a local table

Here's the diagnostic stacktrace (in this case it was a SHOW TABLE STATUS against an
InnoDB table):

stack_bottom = 0x4a6520e8 thread_stack 0x31000
/usr/sbin/mysqld(my_print_stacktrace+0x39)[0x89dca9]
/usr/sbin/mysqld(net_real_write+0x1e7)[0x5b1987]
/usr/sbin/mysqld(net_flush+0x23)[0x5b1e63]
/usr/sbin/mysqld(net_write_command+0x18c)[0x5b201c]
/usr/sbin/mysqld(cli_advanced_command+0xed)[0x72697d]
/usr/sbin/mysqld(mysql_close+0x64)[0x726524]
/usr/sbin/mysqld(_ZN12ha_federated5closeEv+0x29)[0x730179]
/usr/sbin/mysqld(_Z8closefrmP8st_tableb+0x10a)[0x613a6a]
/usr/sbin/mysqld[0x609611]
/usr/sbin/mysqld(my_hash_delete+0x2c0)[0x892b90]
/usr/sbin/mysqld(_Z19close_thread_tablesP3THD+0x18a)[0x60935a]
/usr/sbin/mysqld(_Z14get_all_tablesP3THDP10TABLE_LISTP4Item+0x8ac)[0x6c966c]
/usr/sbin/mysqld(_Z24get_schema_tables_resultP4JOIN23enum_schema_table_state+0x1d1)[0x6c1911]
/usr/sbin/mysqld(_ZN4JOIN4execEv+0x3fe)[0x63853e]
/usr/sbin/mysqld(_Z12mysql_selectP3THDPPP4ItemP10TABLE_LISTjR4ListIS1_ES2_jP8st_orderSB_S2_SB_yP13select_resultP18st_select_lex_unitP13st_select_lex+0x199)[0x63a609]
/usr/sbin/mysqld(_Z13handle_selectP3THDP6st_lexP13select_resultm+0x1c5)[0x63af35]
/usr/sbin/mysqld[0x5c66a1]
/usr/sbin/mysqld(_Z21mysql_execute_commandP3THD+0x3c6)[0x5c9566]
/usr/sbin/mysqld(_Z11mysql_parseP3THDPcjPPKc+0x51b)[0x5cf79b]
/usr/sbin/mysqld(_Z16dispatch_command19enum_server_commandP3THDPcj+0x9f1)[0x5d0191]
/usr/sbin/mysqld(_Z10do_commandP3THD+0x128)[0x5d0dd8]
/usr/sbin/mysqld(handle_one_connection+0x7b1)[0x5c2d91]
/lib64/libpthread.so.0[0x3f15206617]
/lib64/libc.so.6(clone+0x6d)[0x3f14ad3c2d]

Replication is obviously affected by the same problem, so http://bugs.mysql.com/bug.php?id=51196 is likely a
duplicate of this one.

Revision history for this message
Andriy Podanenko (podarokua) wrote :
Revision history for this message
Shahriyar Rzayev (rzayev-sehriyar) wrote :

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-2666

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.