Merge lp:~percona-dev/percona-server/5.1.55-724674 into lp:~percona-dev/percona-server/5.1.55

[Oleg Tsarev]

Patch remove_fcntl_excessive_calls.patch originally is:
http://bazaar.launchpad.net/~mysqlatfacebook/mysqlatfacebook/5.1/revision/3540

Now we update this patch by:
http://bazaar.launchpad.net/~mdcallag/maria/5.3fcntl/revision/2921

[Alexey Kopytov (akopytov)]

Oleg,

Please remove all comment changes from your patch like this one:

+ continue;
+ }
+-#endif
++#endif /* defined(THREAD_SAFE_CLIENT) && !defined(MYSQL_SERVER) */

review: Needs Fixing

[Oleg Tsarev (tsarev)]

Alexey,

What the reason for remove? Some of comments (from original) comment incorrect.

[Oleg Tsarev (tsarev)]

Alexey,

What the reason for remove? Some of comments (from original) commit incorrect.

[Alexey Kopytov (akopytov)]

If by 'original commit' you mean the current state of remove_fcntl_excessive_calls.patch, then I don't see any _changes_ in the comment lines, you are only adding new comments.

If by 'original commit' you mean the MySQL server code, then I don't think you should change anything inb it unless you really have to. In this particular case your goal is to port the patch, not to improve commenting for some randomly selected parts of the server.

[Fred Linhoss (fred-linhoss)]

Documentation changed:

Page changed:
http://www.percona.com/docs/wiki/percona-server:features:remove_fcntl_excessive_calls

Release note added for both 5.1.55-12.6 & 5.5.9-20.1.

review: Approve (documentation)

[Vadim Tkachenko (vadim-tk)]

Please note, it may happen that it won't go into 5.1.55-12.6 but only
in next release.

On Tue, Mar 1, 2011 at 5:43 AM, Fred Linhoss <email address hidden> wrote:
> Review: Approve documentation
> Documentation changed:
>
> Page changed:
> http://www.percona.com/docs/wiki/percona-server:features:remove_fcntl_excessive_calls
>
> Release note added for both 5.1.55-12.6 & 5.5.9-20.1.
>
> --
> https://code.launchpad.net/~percona-dev/percona-server/5.1.55-724674/+merge/51549
> Your team Percona developers is requested to review the proposed merge of lp:~percona-dev/percona-server/5.1.55-724674 into lp:percona-server.
>

--
Vadim Tkachenko, CTO, Percona Inc.
Phone [phone],  Skype: vadimtk153
Schedule meeting: http://tungle.me/VadimTkachenko

[Fred Linhoss (fred-linhoss)]

No additional doc changes made

review: Approve (documentation)

[Alexey Kopytov (akopytov)]

review: Approve

[Oleg Tsarev (tsarev)]

Alexey already review

review: Approve (code review)

[Valentine Gostev (longbow)]

I would like to know what changes this merge introduces, besides the reference to the mentioned revision, is any additional info available?

review: Needs Information (qa)

[Oleg Tsarev (tsarev)]

In:
http://bazaar.launchpad.net/~mysqlatfacebook/mysqlatfacebook/5.1/revision/3540
you can see commit message:

From http://bugs.mysql.com/bug.php?id=54790 I can double peak QPS for
sysbench by removing excessive calls to fcntl in my_real_read. This is a port
of the official fix.

Go to:
http://bugs.mysql.com/bug.php?id=54790

and get information

[Valentine Gostev (longbow)]

review: Approve (qa)

=== modified file 'remove_fcntl_excessive_calls.patch'
--- remove_fcntl_excessive_calls.patch	2011-02-08 23:02:17 +0000
+++ remove_fcntl_excessive_calls.patch	2011-03-02 11:19:45 +0000
@@ -18,6 +18,15 @@
 diff -ruN a/sql/net_serv.cc b/sql/net_serv.cc
 --- a/sql/net_serv.cc	2010-06-03 19:50:27.000000000 +0400
 +++ b/sql/net_serv.cc	2010-07-22 21:40:30.680424001 +0400
+@@ -64,7 +64,7 @@
+   can't normally do this the client should have a bigger max_allowed_packet.
+ */
+ 
+-#if defined(__WIN__) || !defined(MYSQL_SERVER)
++#if (defined(__WIN__) || !defined(MYSQL_SERVER)) && !defined(NO_ALARM)
+   /* The following is because alarms doesn't work on windows. */
+ #define NO_ALARM
+ #endif
 @@ -139,7 +139,7 @@
    if (vio != 0)					/* If real connection */
    {
@@ -36,15 +45,6 @@
        if ((interrupted || length == 0) && !thr_alarm_in_use(&alarmed))
        {
          if (!thr_alarm(&alarmed, net->write_timeout, &alarm_buff))
-@@ -655,7 +655,7 @@
- 		  my_progname);
- #endif /* EXTRA_DEBUG */
-       }
--#if defined(THREAD_SAFE_CLIENT) && !defined(MYSQL_SERVER)
-+#if defined(THREAD_SAFE_CLIENT) && defined(NO_ALARM)
-       if (vio_errno(net->vio) == SOCKET_EINTR)
-       {
- 	DBUG_PRINT("warning",("Interrupted write. Retrying..."));
 @@ -673,7 +673,7 @@
      pos+=length;
      update_statistics(thd_increment_bytes_sent(length));
@@ -71,12 +71,3 @@
  	  /*
  	    We got an error that there was no data on the socket. We now set up
  	    an alarm to not 'read forever', change the socket to non blocking
-@@ -866,7 +867,7 @@
- 		    my_progname,vio_errno(net->vio));
- #endif /* EXTRA_DEBUG */
- 	  }
--#if defined(THREAD_SAFE_CLIENT) && !defined(MYSQL_SERVER)
-+#if defined(THREAD_SAFE_CLIENT) && defined(NO_ALARM)
- 	  if (vio_errno(net->vio) == SOCKET_EINTR)
- 	  {
- 	    DBUG_PRINT("warning",("Interrupted read. Retrying..."));