0001921: 'Main' thread hang if db.query used too much

View Issue Details [ Jump to Notes ]

[ Issue History ] [ Print ]

Project

Category

View Status

Date Submitted

Last Update

0001921

SphereServer

executable - linux build

public

27-02-11 17:46

06-08-14 12:10

Reporter

trinax

Assigned To

XuN

Priority

normal

Severity

major

Reproducibility

sometimes

Status

closed

Resolution

fixed

Platform

Linux

Debian

OS Version

5.0

Product Version

20-09-2009, 0.56b Prerelease

Target Version

Fixed in Version

0.56c Nightly

Summary

0001921: 'Main' thread hang if db.query used too much

Description

I have a script that opens up a dialog that uses db.query when players go into war mode. Sometimes the thing below happens:

16:35:P'Feto' was killed by P'Biscuit'.
16:36:CRITICAL:(event.scp,6787)'Main' thread hang, restarting...

and after the thread restart, logs have the script name on every line:
16:48:(event.scp,6787)World save completed, took 1.0000 seconds

Line 6787 is:
db.query "select name, killamount, uid from the_table where team=2 order by killamount desc";
It is inside a dialog.

Going war mode has a cooldown of a few secs during the event but still this happens when too many people hit tab at the same time I think.

I may be wrong but I think when this happens for the second time without a complete restart, sphere crashes with no error.

Tags

No tags attached.

Nightly Version

Automated (specify build number)

Experimental Flags

DiagonalWalkCheck, New_Triggers, WalkCheck, AllowTelnetPacketFilter, NetworkOutThread

Option Flags

Command_Sysmsgs, FileCommands

Internal Build Number

1420

Attached Files

Relationships

Notes
(0000957) MrSugarCube (administrator) 13-03-11 04:14	What is the value of the FreezeRestartTime setting in sphere.ini? I expect that the cause of this is that there's too much processing going on in the script (and/or the database isn't returning the results fast enough) and Sphere thinks that the process has frozen. Increasing FreezeRestartTime may help. Another option could be to use asynchronous queries (db.aquery/aexecute) if the query is taking a long time to finish.

(0000961) trinax (reporter) 13-03-11 10:00 edited on: 13-03-11 13:18	FreezeRestartTime is 15 seconds. The table usually has 30~rows only. I have just noticed there is two queries in the dialog. Each of them is working with a <local> like this. db.query "select name, k, serial from cs where team=1 order by k desc"; if (<db.row.numrows> > 0) for x 0 <eval <db.row.numrows>-1> local.ct <local.ct><db.row.<dlocal.x>.name> - <db.row.<dlocal.x>.k> <qval <db.row.<dlocal.x>.serial>?- [DEAD]:><def.br> endfor endif db.query "select name, k, serial from cs where team=2 order by k desc"; if (<db.row.numrows> > 0) for x 0 <eval <db.row.numrows>-1> local.ct <local.ct><db.row.<dlocal.x>.name> - <db.row.<dlocal.x>.k> <qval <db.row.<dlocal.x>.serial>?- [DEAD]:><def.br> endfor endif I will try melting the two queries together and see if it is fixed then.

(0002366) XuN (developer) 23-07-14 19:48	If the MySQL server is not on the same machine you should try to use aquery, or you'll have to wait until the data come back. If this is not the case ... is it still happening?

Issue History
Date Modified	Username	Field	Change
27-02-11 17:46	trinax	New Issue
13-03-11 04:14	MrSugarCube	Note Added: 0000957
13-03-11 10:00	trinax	Note Added: 0000961
13-03-11 13:18	trinax	Note Edited: 0000961	View Revisions
23-07-14 19:48	XuN	Note Added: 0002366
06-08-14 12:10	XuN	Status	new => closed
06-08-14 12:10	XuN	Assigned To	=> XuN
06-08-14 12:10	XuN	Resolution	open => fixed
06-08-14 12:10	XuN	Fixed in Version	=> 0.56c Nightly