SC930 log file format

#1

The entries in the session log are of the form <type of
message>:<timestamp1>/<timestamp2>:<other info>, where <timestamp> is
described as being two four-byte integers containing the most accurate
time available in the environment. Here's an example fragment:

SESSION BEGINS(1):1253026078/351875000

DBID=1246951749)(ingres
QUERY:1253026078/351875001:select resolve_table( ~V , ~V )
PARM:1253026078/351875002:21:0=''
PARM:1253026078/351875003:21:1='iisynonyms'
QUERY:1253026078/367500004:define query ~Q is select table_name, table_owner …
PARM:1253026078/367500005:30:0=3088
PARM:1253026078/367500006:30:1=23281
PARM:1253026078/367500007:20:2='ferelacc4
PARM:1253026078/367500008:21:3='ingres'
PARM:1253026078/367500009:21:4='ii%'
PARM:1253026078/367500010:21:5='ingres'
PARM:1253026078/367500011:21:6='ii%'
PARM:1253026078/367500012:21:7='ingres'
ADD-CURSORID:1253026078/367500013

ID=2/4)(ferelacc4 )
EXECUTE:1253026078/383125004

ID=2/4)(ferelacc4 )

It looks pretty obvious (to me) that the low order few bits of the
second part of the timestamp are forced to be distinct and more-or-less
consecutive. But maybe not. Can anyone spare me the trouble of
splashing about in the source code for the answer?

--
Roy

UK Ingres User Association Conference 2010 will be on Tuesday June 8 2010
Go to http://www.iua.org.uk/join to get on the mailing list.

#2

They are theoretically nanoseconds, but on platforms where nanoseconds
were not returned the date routines simply return a sequential number.

I believe the point of adding nanoseconds is to ensure uniqueness. Do
not assuming consecutive entries, though.

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of Roy
Hann
Sent: 16 September 2009 09:43
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: [Info-Ingres] SC930 log file format

The entries in the session log are of the form <type of
message>:<timestamp1>/<timestamp2>:<other info>, where <timestamp> is
described as being two four-byte integers containing the most accurate
time available in the environment. Here's an example fragment:

SESSION BEGINS(1):1253026078/351875000

DBID=1246951749)(ingres
QUERY:1253026078/351875001:select resolve_table( ~V , ~V )
PARM:1253026078/351875002:21:0=''
PARM:1253026078/351875003:21:1='iisynonyms'
QUERY:1253026078/367500004:define query ~Q is select table_name,
table_owner ...
PARM:1253026078/367500005:30:0=3088
PARM:1253026078/367500006:30:1=23281
PARM:1253026078/367500007:20:2='ferelacc4
PARM:1253026078/367500008:21:3='ingres'
PARM:1253026078/367500009:21:4='ii%'
PARM:1253026078/367500010:21:5='ingres'
PARM:1253026078/367500011:21:6='ii%'
PARM:1253026078/367500012:21:7='ingres'
ADD-CURSORID:1253026078/367500013

ID=2/4)(ferelacc4
)
EXECUTE:1253026078/383125004

ID=2/4)(ferelacc4 )

It looks pretty obvious (to me) that the low order few bits of the
second part of the timestamp are forced to be distinct and more-or-less
consecutive. But maybe not. Can anyone spare me the trouble of
splashing about in the source code for the answer?

--
Roy

UK Ingres User Association Conference 2010 will be on Tuesday June 8
2010
Go to http://www.iua.org.uk/join to get on the mailing list.

_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

#3

Robert Kibble wrote:

Quote:

They are theoretically nanoseconds, but on platforms where nanoseconds
were not returned the date routines simply return a sequential number.

Makes sense, good.

Quote:

I believe the point of adding nanoseconds is to ensure uniqueness.

Uniqueness, or strict order?

Quote:

Do not assuming consecutive entries, though.

No, I'm not assuming that; I did say "more-or-less" consecutive.

Many thanks for the answer.

--
Roy

UK Ingres User Association Conference 2010 will be on Tuesday June 8 2010
Go to http://www.iua.org.uk/join to get on the mailing list.

#4

It produces a unique ascending number, strictly in order, at least given
that each sc930 log file is from a single thread. Deffo, as the
youngsters say.

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of Roy
Hann
Sent: 16 September 2009 10:24
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: Re: [Info-Ingres] SC930 log file format

Robert Kibble wrote:

Quote:

They are theoretically nanoseconds, but on platforms where nanoseconds
were not returned the date routines simply return a sequential number.

Makes sense, good.

Quote:

I believe the point of adding nanoseconds is to ensure uniqueness.

Uniqueness, or strict order?

Quote:

Do not assuming consecutive entries, though.

No, I'm not assuming that; I did say "more-or-less" consecutive.

Many thanks for the answer.

--
Roy

UK Ingres User Association Conference 2010 will be on Tuesday June 8
2010
Go to http://www.iua.org.uk/join to get on the mailing list.

_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

#5

Robert Kibble wrote:

Quote:

It produces a unique ascending number, strictly in order, at least given
that each sc930 log file is from a single thread. Deffo, as the
youngsters say.

I have youngsters but the pest-control officer won't touch them.
(Apparently they are "protected" in some way.) Even so, "deffo" is new
to me. :-)

--
Roy

UK Ingres User Association Conference 2010 will be on Tuesday June 8 2010
Go to http://www.iua.org.uk/join to get on the mailing list.

#6

How do I get a SC930 trace log? It looks like a very useful feature!

Roy Hann wrote:

Quote:

The entries in the session log are of the form <type of
message>:<timestamp1>/<timestamp2>:<other info>, where <timestamp> is
described as being two four-byte integers containing the most accurate
time available in the environment. Here's an example fragment:

#7

In tm:

set trace record '/tmp/mydirectory'\p\g
set trace point sc930 1\p\g

/tmp/mydirectory needs to exist.
You will then get trace files appearing in that directory containing
information on everything coming into the dbms.

It is a recent feature, though, so 'set trace record' might not exist if
you're using an old Ingres.

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of
jamesb
Sent: 16 September 2009 11:45
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: Re: [Info-Ingres] SC930 log file format

How do I get a SC930 trace log? It looks like a very useful feature!

Roy Hann wrote:

Quote:

The entries in the session log are of the form <type of
message>:<timestamp1>/<timestamp2>:<other info>, where <timestamp> is
described as being two four-byte integers containing the most accurate
time available in the environment. Here's an example fragment:

_______________________________________________

Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

#8

Oh, and to turn it off again, if you don't want a sorcerer's apprentice
style disk-filling session:

set trace point sc930 0\p\g

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of
Robert Kibble
Sent: 16 September 2009 12:15
To: Ingres and related product discussion forum
Subject: Re: [Info-Ingres] SC930 log file format

In tm:

set trace record '/tmp/mydirectory'\p\g
set trace point sc930 1\p\g

/tmp/mydirectory needs to exist.
You will then get trace files appearing in that directory containing
information on everything coming into the dbms.

It is a recent feature, though, so 'set trace record' might not exist if
you're using an old Ingres.

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of
jamesb
Sent: 16 September 2009 11:45
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: Re: [Info-Ingres] SC930 log file format

How do I get a SC930 trace log? It looks like a very useful feature!

Roy Hann wrote:

Quote:

The entries in the session log are of the form <type of
message>:<timestamp1>/<timestamp2>:<other info>, where <timestamp> is
described as being two four-byte integers containing the most accurate
time available in the environment. Here's an example fragment:

_______________________________________________

Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

#9

Quote:

It is a recent feature, though, so 'set trace record' might not exist
if
you're using an old Ingres.

Is there any documentation which Ingres releases / patch levels have
this feature implemented or not?

Regards
Gerhard

--
gerhard.hofmann (AT) planat (DOT) de
------------------------------------------------------------------------
gerhard.hofmann (AT) planat (DOT) de's Profile: http://community.ingres.com/forum/member.php?userid=857
View this thread: http://community.ingres.com/forum/sh...ad.php?t=11141

#10

If you check the release notes you need SIR 120571 for the initial
release of the sc930 trace point. If requesting a patch, ask for the
fix for bug 122137, which is a placeholder and covers all sc930-related
bugs.

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of
Ingres Forums
Sent: 16 September 2009 15:13
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: Re: [Info-Ingres] SC930 log file format

Quote:

It is a recent feature, though, so 'set trace record' might not exist
if
you're using an old Ingres.

Is there any documentation which Ingres releases / patch levels have
this feature implemented or not?

Regards
Gerhard

--
gerhard.hofmann (AT) planat (DOT) de
------------------------------------------------------------------------
gerhard.hofmann (AT) planat (DOT) de's Profile:
http://community.ingres.com/forum/member.php?userid=857
View this thread:
http://community.ingres.com/forum/sh...ad.php?t=11141

_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres