MySQL crashes constantly and can't get MySQL 5.0.21 to write core-file
Hi, We have major problems with one of our database machines the past few weeks. We've replaced the complete machine about 3 times now, with different memory... no luck. So I doubt it's hardware. It's a SuperMicro SuperServer 6014H-82 ... dual Xeon 3.0, 8GB DDR2-RAM, 4x73GB Maxtor SCSI (RAID10) The symptoms are that MySQL is constantly (eg. within every 30 minutes) getting signal 11's according to the logfile, and is restarting. 060613 08:47:25 mysqld started 060613 8:47:26 InnoDB: Started; log sequence number 0 144990980 060613 8:47:26 [Note] /usr/libexec/mysqld: ready for connections. Version: '5.0.21-debug' socket: '/var/lib/mysql/mysql.sock' port: 3306 Source distribution mysqld got signal 11; This could be because you hit a bug. It is also possible that this binary or one of the libraries it was linked against is corrupt, improperly built, or misconfigured. This error can also be caused by malfunctioning hardware. We will try our best to scrape up some info that will hopefully help diagnose the problem, but since we have already crashed, something is definitely wrong and this may fail. key_buffer_size=67108864 read_buffer_size=6287360 max_used_connections=22 max_connections=750 threads_connected=19 It is possible that mysqld could use up to key_buffer_size + (read_buffer_size + sort_buffer_size)*max_connections = 6206509 K bytes of memory Hope that's ok; if not, decrease some variables in the equation. Writing a core file The memory is OK, it's a machine dedicated to MySQL with 8GB of RAM. We're now running on Fedora Core 5 on x86_64 platform. We were running the latest Fedora 5 RPM's (mysql-5.0.21-2.FC5.1), but since they are compiled with --without-debug I've rebuild it from the SRPM with --with-debug=full When running this, I got no crash... but couldn't run it too long, since the machine became very slow and a lot of clients could no longer connect after sometime due to timeout. When running with --skip-safemalloc I got the crashes as well. I'm running with --core-file and it states 'Writing a core file' in the logs... but it doesn't write it anywhere ! I've also added --core-file-size=100 and even changed system-ulimit for corefiles to unlimited. It just never writes a core file :(( Also tried starting mysqld directly under user mysql, to bypass any possible problem with setuid and not being able to dump core files... no success. If someone could help me with this maybe I'm missing a step :) Thanks in advance ! Regards, -- WideXS http://www.widexs.nl Wouter de Jong System-Administrator Tel +31 (0)23 5698070 Fax +31 (0)23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
Re: MySQL crashes constantly and can't get MySQL 5.0.21 to write core-file
On Tue, Jun 13, 2006 at 08:55:59AM +0200, Wouter de Jong wrote: The symptoms are that MySQL is constantly (eg. within every 30 minutes) getting signal 11's according to the logfile, and is restarting. Problem finally tracked down and filed bug report. Regards, -- WideXS http://www.widexs.nl Wouter de Jong System-Administrator Tel +31 (0)23 5698070 Fax +31 (0)23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
Re: glibc thread_stack
On Mon, Nov 29, 2004 at 11:50:20AM -0700, Sasha Pachev wrote: Hi, Hello, For a few servers with 1000 concurrent connections, I've compiled glibc 2.3.3 with a patched STACK_SIZE in linuxthreads/descr.h : #ifndef STACK_SIZE #define STACK_SIZE (128 * 1024) #endif instead of #define STACK_SIZE (2 * 1024 * 1024) This by the hints on the Documentation @ mysql.com But, now I notice: 041125 15:15:34 mysqld started 041125 15:15:34 Warning: Asked for 196608 thread stack, but got 126976 041125 15:15:35 InnoDB: Started /usr/local/mysql/libexec/mysqld: ready for connections. Is the 128K STACK_SIZE I've built glibc with too little, and should I update it to 256K ? Originally, MySQL team believed 128K was sufficient. Then some time later it was discovered that some DNS resolving routines required a 192K stack in some cases. So for safety reasons, mysqld was modified to request at least 192K stack. In your case, your modification to glibc makes it impossible for it to have a 192K stack, which is why you are getting the message. However, if you run mysqld with --skip-name-resolve (you will need to update your priv tables to use numeric addresses), glibc DNS routines are never called, so 128K should be sufficient. I see :) With MySQL 4.1.8 and higher, I need to use a stack_size of at least 256K, else MySQL segfaults (signal 11) when started. This is with gcc-3.3.5 and glibc-2.3.3. Maybe something you guys should look into this some more, since it really boosts performance :) Also, I have one other question regarding this: When building with --with-other-libc, the build automatically changes to a static build. But, when using OpenSSL, it _must be_ a dynamic build, according to the configure-script. Does anyone have a clue on howto built with a custom glibc, but also have OpenSSL-support ? I really need it :) Thanks ! Kind regards, -- WideXS http://www.widexs.nl Wouter de Jong System-Administrator Tel +31 (0)23 5698070 Fax +31 (0)23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
glibc thread_stack
Hello, For a few servers with 1000 concurrent connections, I've compiled glibc 2.3.3 with a patched STACK_SIZE in linuxthreads/descr.h : #ifndef STACK_SIZE #define STACK_SIZE (128 * 1024) #endif instead of #define STACK_SIZE (2 * 1024 * 1024) This by the hints on the Documentation @ mysql.com But, now I notice: 041125 15:15:34 mysqld started 041125 15:15:34 Warning: Asked for 196608 thread stack, but got 126976 041125 15:15:35 InnoDB: Started /usr/local/mysql/libexec/mysqld: ready for connections. Is the 128K STACK_SIZE I've built glibc with too little, and should I update it to 256K ? max_connections=1500, and max_allowed_packet=25M glibc = 2.3.3 gcc = 3.4.3 mysql = 4.0.22 Please advice :) -- WideXS http://www.widexs.nl Wouter de Jong System-Administrator Tel +31 (0)23 5698070 Fax +31 (0)23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
ERROR 2026: SSL connection error ... what's happening?
Hello, I'm running MySQL 4.0.16 (source compiled) on FreeBSD 5.1, and I have a big problem with SSL... It runs perfectly for a few hours, but then suddenly I can't connect any longer via an SSL-connection... ERROR 2026: SSL connection error That's the error that suddenly appears when using mysql-client, and it also happens with other apllications that use SSL-connection to MySQL Any idea what this could be? ldd: ldd /usr/local/mysql/libexec/mysqld /usr/local/mysql/libexec/mysqld: libssl.so.3 = /usr/lib/libssl.so.3 (0x282bb000) libcrypto.so.3 = /usr/lib/libcrypto.so.3 (0x282ec000) libz.so.2 = /usr/lib/libz.so.2 (0x283f7000) libcrypt.so.2 = /usr/lib/libcrypt.so.2 (0x28405000) libstdc++.so.4 = /usr/lib/libstdc++.so.4 (0x2841e000) libm.so.2 = /usr/lib/libm.so.2 (0x284d3000) libc_r.so.5 = /usr/lib/libc_r.so.5 (0x284f) libc.so.5 = /usr/lib/libc.so.5 (0x28513000) OpenSSL in /usr is OpenSSL 0.9.7a Feb 19 2003 from the base system. -- WideXS http://www.widexs.nl Wouter de Jong System-Administrator Tel +31 (0)23 5698070 Fax +31 (0)23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe:http://lists.mysql.com/[EMAIL PROTECTED]
Performance problems...
may be invalid Number of processes running now: 1 020425 22:20:11 mysqld restarted Should I run with --log to be able to solve this ? Logfile grows to 150MB per 30-45 minutes. (I apologize for this long message ... but I didn't want to forget to mention things :) -- Met vriendelijke groet/With kind regards, Wouter de Jong System-Administrator CABLE WIRELESS Delivering the Internet promise(tm) URL:http://www.widexs.nl Email: [EMAIL PROTECTED] Tel:+31 (0) 23 5698070 Fax:+31 (0) 23 5698099 *** This message may contain information which is confidential or privileged. If you are not the intended recipient, please advise the sender immediately by reply e-mail and delete this message and any attachments without retaining a copy. *** - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: Performance problems...
In addition, here the resolved stack-trace: 0x806cb54 handle_segfault__Fi + 428 0x8116c2a pthread_sighandler + 158 0x80e715e mi_lock_database + 14 0x80b5de8 external_lock__9ha_myisamP3THDi + 28 0x8069d75 lock_external__FPP8st_tableUi + 121 0x8069bea mysql_lock_tables__FP3THDPP8st_tableUi + 362 0x8084e34 lock_tables__FP3THDP13st_table_list + 108 0x808725d open_and_lock_tables__FP3THDP13st_table_list + 41 0x80736ac mysql_execute_command__Fv + 696 0x8077808 mysql_parse__FP3THDPcUi + 216 0x80728f4 do_command__FP3THD + 1460 0x8071ca7 handle_one_connection__FPv + 651 -- Met vriendelijke groet/With kind regards, Wouter de Jong System-Administrator CABLE WIRELESS Delivering the Internet promise(tm) URL:http://www.widexs.nl Email: [EMAIL PROTECTED] Tel:+31 (0) 23 5698070 Fax:+31 (0) 23 5698099 *** This message may contain information which is confidential or privileged. If you are not the intended recipient, please advise the sender immediately by reply e-mail and delete this message and any attachments without retaining a copy. *** - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: Performance problems...
In addition, here the resolved stack-trace: 0x806cb54 handle_segfault__Fi + 428 0x8116c2a pthread_sighandler + 158 0x80e715e mi_lock_database + 14 0x80b5de8 external_lock__9ha_myisamP3THDi + 28 0x8069d75 lock_external__FPP8st_tableUi + 121 0x8069bea mysql_lock_tables__FP3THDPP8st_tableUi + 362 0x8084e34 lock_tables__FP3THDP13st_table_list + 108 0x808725d open_and_lock_tables__FP3THDP13st_table_list + 41 0x80736ac mysql_execute_command__Fv + 696 0x8077808 mysql_parse__FP3THDPcUi + 216 0x80728f4 do_command__FP3THD + 1460 0x8071ca7 handle_one_connection__FPv + 651 Forgot to mention that this stack-trace happened while running 3.23.47 ... the resolved stack-trace is also on the 3.23.47 binary -- Met vriendelijke groet/With kind regards, Wouter de Jong System-Administrator CABLE WIRELESS Delivering the Internet promise(tm) URL:http://www.widexs.nl Email: [EMAIL PROTECTED] Tel:+31 (0) 23 5698070 Fax:+31 (0) 23 5698099 *** This message may contain information which is confidential or privileged. If you are not the intended recipient, please advise the sender immediately by reply e-mail and delete this message and any attachments without retaining a copy. *** - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: renice -20 $$ in safe_mysqld
On Mon, Jan 28, 2002 at 05:30:00PM +0200, Egor Egorov wrote: Hi, In any place that is reached by the scope of execution of safe_mysqld. It must do fine right in the beginning of the script. It doesn't, or I am making a mistake : If I put renice -20 $$ (tried it on several places in safe_mysqld), I get the following: 32606: old priority 0, new priority -20 Starting mysqld daemon with databases from /opt/mysql/var - NICE: nice ---15 nohup -- debug line i added 020130 11:16:20 mysqld ended The .err shows: 020130 11:16:20 mysqld started nice: unrecognized option `---15' Try `nice --help' for more information. 020130 11:16:20 mysqld ended -- Met vriendelijke groet/With kind regards, Wouter de Jong - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: Can't set max_connections on debian version
On Tue, Oct 30, 2001 at 11:31:12AM -0800, William R. Mussatto wrote: Thanks but it doesn't work... Below is the extract of the configuration file it does not matter whether I include spaces around the = or not, starting using mysql_safe results in a failed There is no entry in the error log. Commenting out the line results in a successful restart. I am probably missing something basic but... set-variable= key_buffer=16M set-variable= max_allowed_packet=1M set-variable= thread_stack=128K max_connections = 200 Use this : set-variable= max_connections=200 Since it's a variable : Sincerely, William Mussatto, Senior Systems Engineer CyberStrategies, Inc ph. 909-920-9154 ext. 27 -- Met vriendelijke groet/With kind regards, Wouter de Jong [EMAIL PROTECTED] - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: spontanious performance problem :(
On Tue, Oct 23, 2001 at 08:08:08AM +0200, Michal Zejdl wrote: Blank shot - have you enough FD? sql I really think so... ulimit -a: max user processes 32768 open files 32768 And within /etc/my.cnf: [safe_mysqld] open-files-limit=32768 set-variable= open_files=32768 /proc/sys/fs/file-max: 32768 The load is approx. 0.30 when I don't increase the max_connections from 100 to 1250. So FD was also what I thought about too, but I think all settings for it are OK. -- Wouter de Jong [EMAIL PROTECTED] - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
spontanious performance problem :(
Hi, We are having serious load problems suddenly at one of our MySQL-servers. There are running about 830 databases on this one. The machine was up for 40 days, running peacefully (load avg. of 2) but suddenly it went to an average of 25. We've been trying to find a reason, but SHOW PROCESSLIST didn't tell us much useful info. (No extreme query's, etc). I've run myisamchk on every table, and it fixed a few index problems on some, but no serious stuff (IMHO). The server is a dual PIII/1000, with SCSI-Raid0 and 1024MB RAM. Running MySQL 3.23.40, since upgrading to both .41 and .42 increased the load when we tried it. I see a lot of processes, that are taking about 40 - 90% CPU, and 3 - 10% MEM (RSS 127M sometimes). It was running 2.4.8-ac12, and used a swap of 425MB of 1024MB. I just installed 2.4.12-ac5 on it, but no success (only the swap-usage is 0 again) in terms of improvement. If I do an strace -p process-id on those processes, it just return nothing. So I wonder why those things are eating CPU Mem. Server is running RedHat 6.2, with glibc-2.1.3-22. (gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)) I'm really stuck (again). Last time this happened was about 6 months ago, upgrading to these hardware specs helped a lot. We are very curious about how FreeBSD is performing with SMP and MySQL. (Dual or Quad Xeon procs). Does anyone have a clue about what could be going on ? -- Met vriendelijke groet/With kind regards, Wouter de Jong [EMAIL PROTECTED] - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
linking against mysql-only libpthread.a
Hi, I've read http://www.mysql.com/doc/L/i/Linux.html, and I want to build glibc 2.2.4 on RedHat 6.2 and Slackware 7.1 (currently 2.1.x glibc's). My question : how to build a glibc for MySQL-only, as recommended ? Thus, howto dump it in some directory, and use it in the build process ? Also, do I still need to adjust the STACK_SIZE and PTHREAD_THREADS_MAX ? Another question: 'renice -20 $$' - Who has an example of howto use this correctly ?? -- MadDog2K http://www.maddog2k.nl the great puppy dog - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
RAID advice :
We are planning to use RAID for backing-up MySQL-data. Is it adviced that we put the OS on the RAID-array too, or just the data ? Let's say your OS crashes (Linux...bad libraries for example, that are not resolvable, for example :), then you'll have to format all your disks including your data to replace the OS. Ont he other hand, if the disk that contains your OS crashes, you still have a lot of downtime (well, depends on backup-disk, time it occurs, etc). Anyone a good suggestion ? Or is it even possible to have your OS separate from the data, buth both mirrored (eg. 2 different raid-arrays) -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: Root account cannot access Mysql
On Wed, Jun 13, 2001 at 04:56:23PM -0400, RIDLEY, JAMES A. (AIT) wrote: Hi, I somehow locked out the root account from MySQL. How can I get access back as root on the Unix system? Try this : /path/to/mysql-directory/libexec/mysqld --skip-grant-tables In a second term, just run mysql -p - password: hit enter Works for me... James A Ridley System Administrator AIX ITO Midrange Support Ameritech/SBC Telecommunications Inc. (317)265-4252 work (317)673-3670 pager -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: load-balancing with NFS
On Wed, Jun 13, 2001 at 08:55:50PM +0200, Sven Huster wrote: Hi Hi, As far as i know, DON'T DO THIS mysql with serveral instances accessing the same database will not work as expected. check http://www.mysql.com/doc/M/u/Multiple_servers.html Thanks, I'll skip the plan. No load-balancing for me thus : we run mysql over nfs but only with one instance. seems to be quite stable (db server freebsd - nfs server netapp filer) regards Sven Huster Senior IT Systems Engineer -Original Message- From: Wouter de Jong [mailto:[EMAIL PROTECTED]] Sent: 13 June, 2001 10:21 To: [EMAIL PROTECTED] Subject: load-balancing with NFS *This message was transferred with a trial version of CommuniGate(tm) Pro* Hello, I have an idea about setting up 1 machine that contains the var-directory (RAID5 for backup, etc). Then I mount the var directory from 2 (to explain now :) other servers, via NFS, and run mysqld on those. My question : is this going to work ? I mean, will mysqld allow me to have a shared /var/ directory for all databases inside ? I want load-balancing, cause my server can't handle the pressure good anymore. database-01 IN A 192.168.0.101 database-01 IN A 192.168.0.102 192.168.0.101: (running mysqld) 192.168.0.100:/opt/database/mysql/var /opt/database/mysql/var 192.168.0.102: (running mysqld) 192.168.0.100:/opt/database/mysql/var /opt/database/mysql/var 192.168.0.100: will be NFS-mounted by the 2 mysqld machines And how to configure MySQL with RAID(5) (3 disks, 1 spare) ? Is --with-raid anough, and everything will go automatically good ? I don't have to use other table-types then MyISAM ? Thanks for any input : -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
load-balancing with NFS
Hello, I have an idea about setting up 1 machine that contains the var-directory (RAID5 for backup, etc). Then I mount the var directory from 2 (to explain now :) other servers, via NFS, and run mysqld on those. My question : is this going to work ? I mean, will mysqld allow me to have a shared /var/ directory for all databases inside ? I want load-balancing, cause my server can't handle the pressure good anymore. database-01 IN A 192.168.0.101 database-01 IN A 192.168.0.102 192.168.0.101: (running mysqld) 192.168.0.100:/opt/database/mysql/var /opt/database/mysql/var 192.168.0.102: (running mysqld) 192.168.0.100:/opt/database/mysql/var /opt/database/mysql/var 192.168.0.100: will be NFS-mounted by the 2 mysqld machines And how to configure MySQL with RAID(5) (3 disks, 1 spare) ? Is --with-raid anough, and everything will go automatically good ? I don't have to use other table-types then MyISAM ? Thanks for any input : -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: (sleeping) processes of 90% CPU
On Tue, Jun 12, 2001 at 02:16:27PM +0300, Sinisa Milivojevic wrote: To solve SMP problems with Linux, you should use latest kernel 2.4.* and our binary. We are running 2.4.5-ac4 currently on the server. Could you tell me why your binary should run better ? Cause I tried it, and it dind't make a lot of difference. -- Regards, __ ___ ___ __ / |/ /_ __/ __/ __ \/ /Mr. Sinisa Milivojevic [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, FullTime Developer /_/ /_/\_, /___/\___\_\___/ Larnaca, Cyprus ___/ www.mysql.com -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
Re: (sleeping) processes of 90% CPU
On Sun, Jun 10, 2001 at 01:15:38PM +0300, Sinisa Milivojevic wrote: Hello, I'm a database administrator for a hosting provider, and we have currently 2 SMP MySQL-database servers. The one that is causing me problems, has at least 900 databases running on it. That's an dual PIII-550, with 1024MB memory, running RedHat 6.2, with Linux 2.4.5-ac4. During working hours, a load of 5 6 is pretty default, with about 600 mysqld processes. This is already reduced from 900 to 600, by setting the wait-timeout to 600 seconds, to 120, to 30 at the moment. The problem is that the server is in an unstable situation lately. I see a lot of MySQL-processes with top (sorted by MEM-usage), that are taking over 90% CPU. If I strace -p PID the process, it returns me nothing. I assume this is a sleeping process. RSS sometimes is 120M, most of the time between 20 and 40M. No change in running 3.22.32, 3.23.31 - 3.23.38 (.38 currently). Anyone has a hint on what happens ? SHOW PROCESSLIST gives me for 90% Sleeping as output. Here is (partially) my /etc/my.cnf : skip-locking set-variable= key_buffer=384M set-variable= max_allowed_packet=1M set-variable= table_cache=512 set-variable= sort_buffer=2M set-variable= record_buffer=2M set-variable= thread_cache=8 set-variable= thread_concurrency=4 set-variable= myisam_sort_buffer_size=64M set-variable= max_connections=1250 set-variable= max_connect_errors=9 set-variable= wait_timeout=30 Hi! This is most probably mutex contention problem. Please use our binary, or patch your LinuxThreads with a patch on our site. Hmm, I patched glibc 2.1.3 with the patch on your site, then recompiled MySQL. No change, still huge processes. : I then tried the binary version, that didn't help too. I really don't know what to look for anymore : I can replace the hardware, but I don't think that will change anything... *hoping for hints :* -- Regards, For technical support contracts, go to https://order.mysql.com/ __ ___ ___ __ / |/ /_ __/ __/ __ \/ /Mr. Sinisa Milivojevic [EMAIL PROTECTED] / /|_/ / // /\ \/ /_/ / /__ MySQL AB, FullTime Developer /_/ /_/\_, /___/\___\_\___/ Larnaca, Cyprus ___/ www.mysql.com -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
(sleeping) processes of 90% CPU
Hello, I'm a database administrator for a hosting provider, and we have currently 2 SMP MySQL-database servers. The one that is causing me problems, has at least 900 databases running on it. That's an dual PIII-550, with 1024MB memory, running RedHat 6.2, with Linux 2.4.5-ac4. During working hours, a load of 5 6 is pretty default, with about 600 mysqld processes. This is already reduced from 900 to 600, by setting the wait-timeout to 600 seconds, to 120, to 30 at the moment. The problem is that the server is in an unstable situation lately. I see a lot of MySQL-processes with top (sorted by MEM-usage), that are taking over 90% CPU. If I strace -p PID the process, it returns me nothing. I assume this is a sleeping process. RSS sometimes is 120M, most of the time between 20 and 40M. No change in running 3.22.32, 3.23.31 - 3.23.38 (.38 currently). Anyone has a hint on what happens ? SHOW PROCESSLIST gives me for 90% Sleeping as output. Here is (partially) my /etc/my.cnf : skip-locking set-variable= key_buffer=384M set-variable= max_allowed_packet=1M set-variable= table_cache=512 set-variable= sort_buffer=2M set-variable= record_buffer=2M set-variable= thread_cache=8 set-variable= thread_concurrency=4 set-variable= myisam_sort_buffer_size=64M set-variable= max_connections=1250 set-variable= max_connect_errors=9 set-variable= wait_timeout=30 -- Met vriendelijke groet/With kind regards, Cable Wirelesshttp://www.widexs.nl Wouter de Jong System-Administrator Tel +31 23 5698070 Fax +31 23 5698099 Bijlmermeerstraat 62, 2131 HG HOOFDDORP, NL - Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail [EMAIL PROTECTED] To unsubscribe, e-mail [EMAIL PROTECTED] Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
