Bug #11926 | New Simple Stress test scripts cause MySQLD to crash | ||
---|---|---|---|
Submitted: | 13 Jul 2005 23:22 | Modified: | 22 Sep 2005 16:57 |
Reporter: | Jonathan Miller | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Cluster: Cluster (NDB) storage engine | Severity: | S1 (Critical) |
Version: | 5.1-wl2325 | OS: | Linux (Linux) |
Assigned to: | Tomas Ulin | CPU Architecture: | Any |
[13 Jul 2005 23:22]
Jonathan Miller
[10 Aug 2005 18:38]
Jonathan Miller
This also happens when there is no cluster activity, just using the two scripts with about 5 of each scripts running.
[18 Aug 2005 12:56]
Tomas Ulin
needed to do a create database BANK to get inout.pl to work. Can't repeat with the following: - running all localhost - running 10 of the inout.pl Is there info missing? Should there not be some tables in BANK? What is the significance of the cd_db.pl... it creates a database unrelated to the inout.pl... should be reproducable without this extra process? How long typically before mysql cores? Have done close to 300000 connections without the issue showing up (see below). master> show processlist; +--------+-------------+-----------+---------------------+---------+------+-----------------------------------+----------------------+ | Id | User | Host | db | Command | Time | State | Info | +--------+-------------+-----------+---------------------+---------+------+-----------------------------------+----------------------+ | 1 | system user | | | Daemon | 0 | Waiting for event from ndbcluster | NULL | | 107302 | root | localhost | test | Query | 0 | NULL | show processlist | | 296192 | root | localhost | cluster_replication | Query | 0 | cleaning up | DROP DATABASE TESTER | | 296194 | root | localhost | BANK | Quit | 0 | cleaning up | NULL | | 296196 | root | localhost | BANK | Quit | 0 | cleaning up | NULL | | 296199 | root | localhost | BANK | Sleep | 0 | NULL | NULL | | 296202 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296204 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296205 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296206 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296207 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296208 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296209 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296210 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296211 | root | localhost | BANK | Quit | 0 | NULL | NULL | | 296212 | root | localhost | NULL | Connect | NULL | login | NULL | | 296213 | root | localhost | NULL | Connect | NULL | login | NULL | | 296214 | root | localhost | NULL | Connect | NULL | login | NULL | | 296215 | root | localhost | NULL | Connect | NULL | login | NULL | | 296216 | root | localhost | NULL | Connect | NULL | login | NULL | | 296217 | root | localhost | NULL | Connect | NULL | login | NULL | | 296218 | root | localhost | NULL | Connect | NULL | login | NULL | | 296219 | root | localhost | NULL | Connect | NULL | login | NULL | | 296220 | root | localhost | NULL | Connect | NULL | login | NULL | | 296221 | root | localhost | NULL | Connect | NULL | login | NULL | | 296222 | root | localhost | NULL | Connect | NULL | login | NULL | +--------+-------------+-----------+---------------------+---------+------+-----------------------------------+----------------------+ 26 rows in set (0.01 sec)
[18 Aug 2005 13:19]
Jonathan Miller
needed to do a create database BANK to get inout.pl to work. or *you can just change the script to use test. Should there not be some tables in BANK? *Not needed What is the significance of the cd_db.pl... it creates a database unrelated to the inout.pl... should be reproducable without this extra process? *It adds stress, I.E the purpose of stress testing it to add stress to the system. Note: 2 sessions of BANK, 2 sessions of Sabre were also being done. Also, stack trace was provided: 0x816e55c 0x4005e5cd 0x835c2f7 0x835b6fe 0x834ad6a 0x8349bad 0x8349bad 0x838fd4e 0x8398930 0x83972a2 0x8390203 0x839017f 0x83805f5 0x400586de 0x401d86c7
[22 Sep 2005 16:57]
Jonathan Miller
We are no longer able to reproduce the core associated with this bug report with the latest version of the 5.1-wl2325.