4.0.20 replication

am 15.06.2004 16:59:15 von dephilli

Re: 4.0.20 replication

am 15.06.2004 17:46:44 von Sinisa Milivojevic

DePhillips, Michael P writes:
> Hello,
>
>
>
> My 4.0.20 master is not releasing threads created to do a binlog dump.
>
>
>
> These accumulate over time.
>
>
>
> For example
>
>
>
> Show processlist (from master)
>
>

[skip]
>
>
>
> (x's replaced domain for security reasons.)
>
>
>
>
>
> Shouldn't a replication thread be released before a new one is created?
>
>
>
> Thanks
>
> Michael
>

And how many slaves do you have ??

--

Sincerely,

--
For technical support contracts, go to https://order.mysql.com/?ref=msmi
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB
/_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
<___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 15.06.2004 19:21:22 von Sinisa Milivojevic

DePhillips, Michael P writes:
> Hello
>
> I have 10 slaves, as you can see from the output - there is a connection to
> each, a binlog dump is completed (successfully) and then a new connection is
> made, the old connection(s) remain(s).
>
>
> Thank you for looking into this.
>
> Michael

And how many binlog dumping connections are there ???

And are you using our binary ??

--

Sincerely,

--
For technical support contracts, go to https://order.mysql.com/?ref=msmi
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB
/_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
<___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 15.06.2004 19:44:17 von dephilli

Hello

I have 10 slaves, as you can see from the output - there is a connection to
each, a binlog dump is completed (successfully) and then a new connection is
made, the old connection(s) remain(s).

Thank you for looking into this.

Michael
> -----Original Message-----
> From: Sinisa Milivojevic [mailto:sinisa@mysql.com]
> Sent: Tuesday, June 15, 2004 11:47 AM
> To: DePhillips, Michael P
> Cc: bugs@lists.mysql.com
> Subject: Re: 4.0.20 replication
>
> DePhillips, Michael P writes:
> > Hello,
> >
> >
> >
> > My 4.0.20 master is not releasing threads created to do a binlog dump.
> >
> >
> >
> > These accumulate over time.
> >
> >
> >
> > For example
> >
> >
> >
> > Show processlist (from master)
> >
> >
>
> [skip]
> >
> >
> >
> > (x's replaced domain for security reasons.)
> >
> >
> >
> >
> >
> > Shouldn't a replication thread be released before a new one is created?
> >
> >
> >
> > Thanks
> >
> > Michael
> >
>
> And how many slaves do you have ??
>
> --
>
> Sincerely,
>
> --
> For technical support contracts, go to https://order.mysql.com/?ref=msmi
> __ ___ ___ ____ __
> / |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
> / /|_/ / // /\ \/ /_/ / /__ MySQL AB
> /_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
> <___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 15.06.2004 20:40:09 von dephilli

> And how many binlog dumping connections are there ???

As of now there are ~110 and this number is increasing over time. Last
night when I restarted the (master) server @~2:00 am EST there were ~10 as
expected then slowly the binlog connections began to increase. This will
continue until I run out of connections again - (which, of course will be at
2:00 am) or until I restart the server.

> And are you using our binary ??
yes - mysql-standard-4.0.20-pc-linux-i686.tar.gz

> -----Original Message-----
> From: Sinisa Milivojevic [mailto:sinisa@mysql.com]
> Sent: Tuesday, June 15, 2004 1:21 PM
> To: DePhillips, Michael P
> Cc: bugs@lists.mysql.com
> Subject: RE: 4.0.20 replication
>
> DePhillips, Michael P writes:
> > Hello
> >
> > I have 10 slaves, as you can see from the output - there is a connection
> to
> > each, a binlog dump is completed (successfully) and then a new
> connection is
> > made, the old connection(s) remain(s).
> >
> >
> > Thank you for looking into this.
> >
> > Michael
>
> And how many binlog dumping connections are there ???
>
> And are you using our binary ??
>
> --
>
> Sincerely,
>
> --
> For technical support contracts, go to https://order.mysql.com/?ref=msmi
> __ ___ ___ ____ __
> / |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
> / /|_/ / // /\ \/ /_/ / /__ MySQL AB
> /_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
> <___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 16.06.2004 11:36:02 von Sinisa Milivojevic

DePhillips, Michael P writes:
> > And how many binlog dumping connections are there ???
>
> As of now there are ~110 and this number is increasing over time. Last
> night when I restarted the (master) server @~2:00 am EST there were ~10 as
> expected then slowly the binlog connections began to increase. This will
> continue until I run out of connections again - (which, of course will be at
> 2:00 am) or until I restart the server.
>
> > And are you using our binary ??
> yes - mysql-standard-4.0.20-pc-linux-i686.tar.gz
>
>
>

Seems like we shall need to take a deeper look at this.

Can you send us your master my.cnf and tell us what to do exactly to
get 110 or more connections for 10 slaves ??

--

Sincerely,

--
For technical support contracts, go to https://order.mysql.com/?ref=msmi
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB
/_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
<___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 16.06.2004 17:51:10 von dephilli

Hello Sinisa

> Seems like we shall need to take a deeper look at this.
OK.

> Can you send us your master my.cnf

# The following options will be passed to all MySQL clients
[client]
port = 3306
socket = /tmp/mysql.3306.sock

# Here follows entries for some specific programs

# The MySQL server
[mysqld]
port = 3306
socket = /tmp/mysql.3306.sock
datadir = /db01/offline/data
skip-locking
set-variable = key_buffer=256M
set-variable = max_allowed_packet=1M
set-variable = table_cache=1024
set-variable = max_connections=512
set-variable = sort_buffer=1M
set-variable = record_buffer=1M
set-variable = myisam_sort_buffer_size=64M
set-variable = thread_cache=8
set-variable = thread_concurrency=8 # Try number of CPU's*2
log-bin
server-id = 880673316

[mysqldump]
quick
set-variable = max_allowed_packet=16M

[mysql]
no-auto-rehash

>tell us what to do exactly to get 110 or more connections for 10 slaves ??
BTW. It will increase to 512 if I let it (see max-connections above)

The machine its running on:
uname -a Linux robinson.star.bnl.gov 2.4.20-30.7.legacysmp #1 SMP Fri Feb 20
10:12:55 PST 2004 i686 unknown

The slaves are on similar machines running your binaries of versions ranging
from 4.0.16 to 4.0.20.

I've been running this type of replication for years, and with no problems.
I've done nothing different or special in setting this up. It is pretty
much done right our of Paul's book.

One point you will be interested in...
I've recently upgraded the master from 3.23.58 to 4.0.20, this is when the
problem began. All my slaves where 4.0.xx for over a year now. (yes I ran
mysql_fix_privilege_tables).

Replication is being done. Show master status corresponds to show slave
status etc. The connections, however, keep growing.

I gladly supply any information I can and help in any way possible. Also I
was going to drop back a version on the master, but I'll keep this one
running as long as I know your working on the problem.

Thanks
Michael

> -----Original Message-----
> From: Sinisa Milivojevic [mailto:sinisa@mysql.com]
> Sent: Wednesday, June 16, 2004 5:36 AM
> To: DePhillips, Michael P
> Cc: bugs@lists.mysql.com
> Subject: RE: 4.0.20 replication
>
> DePhillips, Michael P writes:
> > > And how many binlog dumping connections are there ???
> >
> > As of now there are ~110 and this number is increasing over time. Last
> > night when I restarted the (master) server @~2:00 am EST there were ~10
> as
> > expected then slowly the binlog connections began to increase. This
> will
> > continue until I run out of connections again - (which, of course will
> be at
> > 2:00 am) or until I restart the server.
> >
> > > And are you using our binary ??
> > yes - mysql-standard-4.0.20-pc-linux-i686.tar.gz
> >
> >
> >
>
> Seems like we shall need to take a deeper look at this.
>
> Can you send us your master my.cnf and tell us what to do exactly to
> get 110 or more connections for 10 slaves ??
>
> --
>
> Sincerely,
>
> --
> For technical support contracts, go to https://order.mysql.com/?ref=msmi
> __ ___ ___ ____ __
> / |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
> / /|_/ / // /\ \/ /_/ / /__ MySQL AB
> /_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
> <___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 16.06.2004 18:19:57 von Sinisa Milivojevic

DePhillips, Michael P writes:
> Hello Sinisa
>
> > Seems like we shall need to take a deeper look at this.
> OK.
>
> > Can you send us your master my.cnf
>
> # The following options will be passed to all MySQL clients
> [client]
> port = 3306
> socket = /tmp/mysql.3306.sock
>
> # Here follows entries for some specific programs
>
> # The MySQL server
> [mysqld]
> port = 3306
> socket = /tmp/mysql.3306.sock
> datadir = /db01/offline/data
> skip-locking
> set-variable = key_buffer=256M
> set-variable = max_allowed_packet=1M
> set-variable = table_cache=1024
> set-variable = max_connections=512
> set-variable = sort_buffer=1M
> set-variable = record_buffer=1M
> set-variable = myisam_sort_buffer_size=64M
> set-variable = thread_cache=8
> set-variable = thread_concurrency=8 # Try number of CPU's*2
> log-bin
> server-id = 880673316
>
> [mysqldump]
> quick
> set-variable = max_allowed_packet=16M
>
> [mysql]
> no-auto-rehash
>
>
> >tell us what to do exactly to get 110 or more connections for 10 slaves ??
> BTW. It will increase to 512 if I let it (see max-connections above)
>
> The machine its running on:
> uname -a Linux robinson.star.bnl.gov 2.4.20-30.7.legacysmp #1 SMP Fri Feb 20
> 10:12:55 PST 2004 i686 unknown
>
> The slaves are on similar machines running your binaries of versions ranging
> from 4.0.16 to 4.0.20.
>
> I've been running this type of replication for years, and with no problems.
> I've done nothing different or special in setting this up. It is pretty
> much done right our of Paul's book.
>
> One point you will be interested in...
> I've recently upgraded the master from 3.23.58 to 4.0.20, this is when the
> problem began. All my slaves where 4.0.xx for over a year now. (yes I ran
> mysql_fix_privilege_tables).
>
> Replication is being done. Show master status corresponds to show slave
> status etc. The connections, however, keep growing.
>
> I gladly supply any information I can and help in any way possible. Also I
> was going to drop back a version on the master, but I'll keep this one
> running as long as I know your working on the problem.
>
> Thanks
> Michael
>

Thanks for the info provided.

We shall try to repeat the behaviour.

--

Sincerely,

--
For technical support contracts, go to https://order.mysql.com/?ref=msmi
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB
/_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
<___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 16.06.2004 18:24:12 von Sinisa Milivojevic

DePhillips, Michael P writes:
> Hello Sinisa
>
> > Seems like we shall need to take a deeper look at this.
> OK.
>
> > Can you send us your master my.cnf
>
> # The following options will be passed to all MySQL clients
> [client]
> port = 3306
> socket = /tmp/mysql.3306.sock
>
> # Here follows entries for some specific programs
>
> # The MySQL server
> [mysqld]
> port = 3306
> socket = /tmp/mysql.3306.sock
> datadir = /db01/offline/data
> skip-locking
> set-variable = key_buffer=256M
> set-variable = max_allowed_packet=1M
> set-variable = table_cache=1024
> set-variable = max_connections=512
> set-variable = sort_buffer=1M
> set-variable = record_buffer=1M
> set-variable = myisam_sort_buffer_size=64M
> set-variable = thread_cache=8
> set-variable = thread_concurrency=8 # Try number of CPU's*2
> log-bin
> server-id = 880673316
>
> [mysqldump]
> quick
> set-variable = max_allowed_packet=16M
>
> [mysql]
> no-auto-rehash
>
>
> >tell us what to do exactly to get 110 or more connections for 10 slaves ??
> BTW. It will increase to 512 if I let it (see max-connections above)
>
> The machine its running on:
> uname -a Linux robinson.star.bnl.gov 2.4.20-30.7.legacysmp #1 SMP Fri Feb 20
> 10:12:55 PST 2004 i686 unknown
>
> The slaves are on similar machines running your binaries of versions ranging
> from 4.0.16 to 4.0.20.
>
> I've been running this type of replication for years, and with no problems.
> I've done nothing different or special in setting this up. It is pretty
> much done right our of Paul's book.
>
> One point you will be interested in...
> I've recently upgraded the master from 3.23.58 to 4.0.20, this is when the
> problem began. All my slaves where 4.0.xx for over a year now. (yes I ran
> mysql_fix_privilege_tables).
>
> Replication is being done. Show master status corresponds to show slave
> status etc. The connections, however, keep growing.
>
> I gladly supply any information I can and help in any way possible. Also I
> was going to drop back a version on the master, but I'll keep this one
> running as long as I know your working on the problem.
>
> Thanks
> Michael
>
>

Hi!

We do need some more info.

Just check that server-id is different on all slaves and dffers from
the master.

Also, send us slave info for 3 or 4 of your slaves.

We need replication user , it's privileges and output from SHOW SLAVE
STATUS from the same 3 or 4 slave installations.

--

Sincerely,

--
For technical support contracts, go to https://order.mysql.com/?ref=msmi
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB
/_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
<___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org

RE: 4.0.20 replication

am 18.06.2004 15:42:42 von Sinisa Milivojevic

> Hi -
>
> The serverIDs are the _same_ on all the slaves but they (as a group) differ
> from the master.

ok, that explains it.
Server ids must be different on all slaves of the same master.
Otherwise, each time a slave connects to the master, it will stop the
replication of all other threads; because master will believe that the
same slave is connecting twice (or more), whereas only one connection
is needed by slave, so it will kill all "useless" connections.
Will emphasize it in the manual.

Please keep in mind that this list is dedicated to true bugs with
repeatable test cases.

It is not to be used for free support, like in this case.

--

Sincerely,

--
For technical support contracts, go to https://order.mysql.com/?ref=msmi
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Mr. Sinisa Milivojevic
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB
/_/ /_/\_, /___/\___\_\___/ Full time Developer and Support Coordinator
<___/ www.mysql.com Larnaca, Cyprus

--
MySQL Bugs Mailing List
For list archives: http://lists.mysql.com/bugs
To unsubscribe: http://lists.mysql.com/bugs?unsub=gcdmb-bugs@m.gmane.org