Comment 5 for bug 643650

Revision history for this message
Vadim Tkachenko (vadim-tk) wrote : Re: [Bug 643650] [NEW] InnoDB with shared buffer pool takes long to shutdown

Yasufumi,

We can use new checksum from Facebook, they have implementation based
on CPU instructions.

Why do you need to check checksum at shutdown?

I am setting this as "high" because the purpose of feature is to
minimize downtime, and this bug makes it problematic.

On Tuesday, September 21, 2010, Yasufumi Kinoshita
<email address hidden> wrote:
> Anyway, calc check sum is needed to confirm destruction, I think.
> It should be changed to ut_fold_binary_32 of innodb_fast_checksum.patch?
> But it might make around 1/2 of the time.
>
> And why the importance of it is "High"?
> It doesn't break anything data. Not danger.
> No magic solutions.
>
> --
> InnoDB with shared buffer pool takes long to shutdown
> https://bugs.launchpad.net/bugs/643650
> You received this bug notification because you are a member of Percona
> developers, which is the registrant for Percona Server.
>
> Status in Percona Server with XtraDB: Confirmed
>
> Bug description:
> I noticed that on server with big amount of memory ( 120GB buffer pool)
> when I use buffer_pool_shared memory it takes very long to shutdown.
>
> There is some process after InnoDB finished flushing, it is CPU bound process.
>
> stack trace:
>
>
> (gdb) thread app all bt
>
> Thread 2 (Thread 0x599e2940 (LWP 17830)):
> #0  buf_pool_free () at ./include/ut0rnd.ic:224
> #1  0x0000000000798a3a in innobase_shutdown_for_mysql () at srv/srv0start.c:2215
> #2  0x000000000072c78f in innobase_end (hton=<value optimized out>, type=<value optimized out>) at handler/ha_innodb.cc:2588
> #3  0x000000000068f6ab in ha_finalize_handlerton (plugin=0x5725db8) at handler.cc:395
> #4  0x000000000070cbc6 in plugin_deinitialize (plugin=0x5725db8, ref_check=true) at sql_plugin.cc:828
> #5  0x000000000070d6fe in reap_plugins () at sql_plugin.cc:916
> #6  0x000000000070db8d in plugin_shutdown () at sql_plugin.cc:1551
> #7  0x00000000005b01bc in clean_up (print_message=true) at mysqld.cc:1363
> #8  0x00000000005b061e in unireg_end () at mysqld.cc:1289
> #9  0x00000000005b430d in kill_server (arg=<value optimized out>) at mysqld.cc:1224
> #10 kill_server_thread (arg=<value optimized out>) at mysqld.cc:1252
> #11 0x0000003d2760673d in start_thread () from /lib64/libpthread.so.0
> #12 0x0000003d26ed3d1d in clone () from /lib64/libc.so.6
>
> Thread 1 (Thread 0x2ab90c584fb0 (LWP 17698)):
> #0  0x0000003d2760aee9 in pthread_cond_wait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0
> #1  0x00000000005b52f5 in main (argc=<value optimized out>, argv=<value optimized out>) at mysqld.cc:4765
>
>
>

--
Vadim Tkachenko, CTO, Percona Inc.
Phone +1-888-401-3403, Skype: vadimtk153
Schedule meeting: http://tungle.me/VadimTkachenko