From prwolfe at sandia.gov  Mon May  8 12:56:22 2017
From: prwolfe at sandia.gov (Wolfenbarger, Paul R)
Date: Mon, 8 May 2017 16:56:22 +0000
Subject: [CDash] [EXTERNAL] RE: slow label query
In-Reply-To: <0863544ce310402a9d7b33b3f4f86cac@azdexchstore3.AZD.LOCAL>
References: <FFC366FE-4E18-4D95-BDC9-0E327217A406@sandia.gov>
 <0863544ce310402a9d7b33b3f4f86cac@azdexchstore3.AZD.LOCAL>
Message-ID: <D3AC061C-89EA-419A-A06B-658BF6DA43A5@sandia.gov>

I found some time and set up a copy of our installation, modified the build2test table as referenced in your mail thread and noted 2 things


1.       I stopped getting duplicate entries in this table.  Since our size is in the 10?s of millions anyway that is a welcome change.

2.       Query times are not changed.  Running explain on that query still shows a write to a temporary table.

I will get a pull request together as I see this change as somewhat helpful, but it does not solve my initial problem

----------------------------------------------------------------------------------------
At the heart of science is an essential tension between two seemingly
contradictory attitudes -- an openness to new ideas, no matter how bizarre
or counterintuitive they may be, and the most ruthless skeptical scrutiny
of all ideas, old and new.  This is how deep truths are winnowed from deep
nonsense.  Of course, scientists make mistakes in trying to understand the
world, but there is a built-in error-correcting mechanism:  The collective
enterprise of creative thinking and skeptical thinking together keeps the
field on track.
-- Carl Sagan, "The Fine Art of Baloney Detection"

Paul R. Wolfenbarger, P.E., M.S.
Computational Simulation Infrastructure
DevOps Product Owner
Sandia National Laboratories
P.O. Box 5800 MS 0845
Albuquerque, NM  87185-0845
505-844-5458 phone
prwolfe at sandia.gov<mailto:prwolfe at sandia.gov>
----------------------------------------------------------------------------------------

From: "Mat?j? Miroslav, Ing." <Mateju.Miroslav at azd.cz>
Date: Monday, February 27, 2017 at 12:31 AM
To: "Wolfenbarger, Paul R" <prwolfe at sandia.gov>
Cc: "cdash at public.kitware.com" <cdash at public.kitware.com>
Subject: [EXTERNAL] RE: slow label query

Hi Paul,

As far as I know (but I am rather a C++ programmer like you than a DB expert), it is better to let MySQL do as much as possible and avoid the overhead of transfers and conversions between MySQL and PHP.

I think your issue is related to my problem with build2test table performance that I described in the mail thread Optimizing build2test table<http://public.kitware.com/pipermail/cdash/2016-August/001692.html>. Unfortunately, I haven?t been able to test and push the solution yet but it is quite easy to implement in an existing installation: Just add the testid column as the second column to the buildid index for the build2test table. I have done it using PhpMyAdmin on my production machine. It took about 2 hours (for more than 100 million rows) and the submissions (which were the main issue for me) begun to be processed about 3 times faster after that.

Hope this helps,

Ing. Miroslav Mat?j?
Programmer Analyst

A?D Praha s.r.o.
Technology Division
Research and Development
?irovnick? 2/3146, 106 17  Prague
Czech Republic
Phone: +420 267 287 476
Web: www.azd.cz<http://www.azd.cz/>

From: CDash [mailto:cdash-bounces at public.kitware.com] On Behalf Of Wolfenbarger, Paul R
Sent: Thursday, February 23, 2017 5:35 PM
To: cdash at public.kitware.com
Subject: [CDash] slow label query

I am hoping for someone with more sql understanding than I have to help with an issue. In api/v1/index.php at line 1004 there is a label query that is causing a bottleneck in my installation

                $label_query_base =
                    "SELECT b2t.status, b2t.newstatus
                    FROM build2test AS b2t
                    INNER JOIN label2test AS l2t ON
                    (l2t.testid=b2t.testid AND l2t.buildid=b2t.buildid)
                    WHERE b2t.buildid = '$buildid' AND
                    l2t.labelid IN $label_ids";
                $label_filter_query = $label_query_base . $limit_sql;
                $labels_result = pdo_query($label_filter_query);

This is returning two non-indexed fields from build2test and joining on indexed fields and since my build2test and label2test tables are each about 43 million records, this takes 20 to 300 seconds each. Multipled by at least 40 builds in the loop and you can see my issue.

As a c++ programmer my first instinct was to split the query and make all the operations work on indexes. I did some queries at the command line and got sub-second returns from each query and the php would now look something like

                $testids = pdo_query(
                  ?select testid from label2test
                   where label2test.buildid=$buildid
                   and label2test.labelid in $label_ids") ;
                $test_ids = implode("','", $testids) ;
                $label_filter_query =
                  "select status, newstatus
                   from build2test where buildid = $buildid
                   and testid in $test_ids" ;

However, from my reading it seems that multiple queries of this sort are frowned on in the sql world.  Before I set up a test instance to check this out do any of you have any insight into how to re-structure the query to get the kind of performance I need?

Thanks!

----------------------------------------------------------------------------------------
"If you understand what you're doing, you're not learning anything."
  -- A. L.

Paul R. Wolfenbarger, P.E., M.S.
Computational Simulation Infrastructure
DevOps Product Owner
Sandia National Laboratories
P.O. Box 5800 MS 0845
Albuquerque, NM  87185-0845
505-844-5458 phone
prwolfe at sandia.gov<mailto:prwolfe at sandia.gov>
----------------------------------------------------------------------------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/cdash/attachments/20170508/adf2050e/attachment-0001.html>

From daibane at sandia.gov  Fri May 19 13:35:16 2017
From: daibane at sandia.gov (Ibanez, Daniel Alejandro)
Date: Fri, 19 May 2017 17:35:16 +0000
Subject: [CDash] Reducing upload size
Message-ID: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>

Hello,

We operate the following site:

http://my.cdash.org/index.php?project=Albany

It seems that we?re exceeding the upload quota.
I?m also told that the paid subscription plans are going away or gone.
So the first question is:

In the long-term is it a good idea for us to plan on hosting our own CDash instance? Will my.cdash.org be around much longer?

We control the whole process using the scripting feature of CTest and a fairly elaborate set of scripts, for example see this important file:

https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake

Ideally, we would like to reduce the amount of information uploaded. My guess is the vast majority of the data is in test output. I know there is a way to stop CTest from printing test output to the console unless the test actually fails, is there an equivalent option that affects the uploaded XML file?

Thanks,
Dan

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/cdash/attachments/20170519/4fd1dc93/attachment.html>

From DLRdave at aol.com  Fri May 19 15:15:48 2017
From: DLRdave at aol.com (David Cole)
Date: Fri, 19 May 2017 15:15:48 -0400
Subject: [CDash] Reducing upload size
In-Reply-To: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
Message-ID: <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>

You can set the variable
https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
in a CTestCustom file to limit the amount of output which is recorded
in the uploaded XML files for passed tests. (Typically this output is
the same from run to run, and does not need to be reported to the
dashboard.)

Since your project uses the stuff from Trilinos, I suspect there's a
setting in there somewhere which sets this very high to make it easy
for your devs to see the test output generated by "other computers".
If you can find a way to set it in your project's CTestCustom file(s)
then you should be able to reduce your uploaded volume of data
significantly.

I think the default value for this parameter is something like 1k.
(You could check the CMake/CTest source code to verify.)


HTH,
David C.



On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
<daibane at sandia.gov> wrote:
> Hello,
>
>
>
> We operate the following site:
>
>
>
> http://my.cdash.org/index.php?project=Albany
>
>
>
> It seems that we?re exceeding the upload quota.
>
> I?m also told that the paid subscription plans are going away or gone.
>
> So the first question is:
>
>
>
> In the long-term is it a good idea for us to plan on hosting our own CDash
> instance? Will my.cdash.org be around much longer?
>
>
>
> We control the whole process using the scripting feature of CTest and a
> fairly elaborate set of scripts, for example see this important file:
>
>
>
> https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
>
>
>
> Ideally, we would like to reduce the amount of information uploaded. My
> guess is the vast majority of the data is in test output. I know there is a
> way to stop CTest from printing test output to the console unless the test
> actually fails, is there an equivalent option that affects the uploaded XML
> file?
>
>
>
> Thanks,
>
> Dan
>
>
>
>
> _______________________________________________
> CDash mailing list
> CDash at public.kitware.com
> http://public.kitware.com/mailman/listinfo/cdash
>

From daibane at sandia.gov  Fri May 19 15:22:30 2017
From: daibane at sandia.gov (Ibanez, Daniel Alejandro)
Date: Fri, 19 May 2017 19:22:30 +0000
Subject: [CDash] [EXTERNAL] Re:  Reducing upload size
In-Reply-To: <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
 <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
Message-ID: <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>

Thanks, David.

We have full control of how to do this, so that?s good.
I?m not familiar with CTestCustom (as opposed to CTestConfig),
but here is my best guess at usage:

1. Create CTestCustom.cmake, put it next to the CTest script.
2. In the CTest script, do

configure_file(${CMAKE_CURRENT_LIST_DIR}/CTestCustom.cmake ${CTEST_BINARY_DIRECTORY}/CTestCustom.cmake COPYONLY)

prior to calling ctest_start().

Does that sound right?
I?ll give it a try.

Dan

On 5/19/17, 1:15 PM, "David Cole" <DLRdave at aol.com> wrote:

    You can set the variable
    https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
    in a CTestCustom file to limit the amount of output which is recorded
    in the uploaded XML files for passed tests. (Typically this output is
    the same from run to run, and does not need to be reported to the
    dashboard.)
    
    Since your project uses the stuff from Trilinos, I suspect there's a
    setting in there somewhere which sets this very high to make it easy
    for your devs to see the test output generated by "other computers".
    If you can find a way to set it in your project's CTestCustom file(s)
    then you should be able to reduce your uploaded volume of data
    significantly.
    
    I think the default value for this parameter is something like 1k.
    (You could check the CMake/CTest source code to verify.)
    
    
    HTH,
    David C.
    
    
    
    On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
    <daibane at sandia.gov> wrote:
    > Hello,
    >
    >
    >
    > We operate the following site:
    >
    >
    >
    > http://my.cdash.org/index.php?project=Albany
    >
    >
    >
    > It seems that we?re exceeding the upload quota.
    >
    > I?m also told that the paid subscription plans are going away or gone.
    >
    > So the first question is:
    >
    >
    >
    > In the long-term is it a good idea for us to plan on hosting our own CDash
    > instance? Will my.cdash.org be around much longer?
    >
    >
    >
    > We control the whole process using the scripting feature of CTest and a
    > fairly elaborate set of scripts, for example see this important file:
    >
    >
    >
    > https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
    >
    >
    >
    > Ideally, we would like to reduce the amount of information uploaded. My
    > guess is the vast majority of the data is in test output. I know there is a
    > way to stop CTest from printing test output to the console unless the test
    > actually fails, is there an equivalent option that affects the uploaded XML
    > file?
    >
    >
    >
    > Thanks,
    >
    > Dan
    >
    >
    >
    >
    > _______________________________________________
    > CDash mailing list
    > CDash at public.kitware.com
    > http://public.kitware.com/mailman/listinfo/cdash
    >
    


From DLRdave at aol.com  Fri May 19 15:30:14 2017
From: DLRdave at aol.com (David Cole)
Date: Fri, 19 May 2017 15:30:14 -0400
Subject: [CDash] [EXTERNAL] Re:  Reducing upload size
In-Reply-To: <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
 <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
 <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>
Message-ID: <CAKME8fFF1twW_2Ep2m0HGqiXzco+7i2aCutUST2SJsXPxGxJfA@mail.gmail.com>

Sounds like it will almost work.

You'll also likely need a call to
https://cmake.org/cmake/help/latest/command/ctest_read_custom_files.html#command:ctest_read_custom_files
in your ctest script after the configure step.


Let us know if it works out,
David



On Fri, May 19, 2017 at 3:22 PM, Ibanez, Daniel Alejandro
<daibane at sandia.gov> wrote:
> Thanks, David.
>
> We have full control of how to do this, so that?s good.
> I?m not familiar with CTestCustom (as opposed to CTestConfig),
> but here is my best guess at usage:
>
> 1. Create CTestCustom.cmake, put it next to the CTest script.
> 2. In the CTest script, do
>
> configure_file(${CMAKE_CURRENT_LIST_DIR}/CTestCustom.cmake ${CTEST_BINARY_DIRECTORY}/CTestCustom.cmake COPYONLY)
>
> prior to calling ctest_start().
>
> Does that sound right?
> I?ll give it a try.
>
> Dan
>
> On 5/19/17, 1:15 PM, "David Cole" <DLRdave at aol.com> wrote:
>
>     You can set the variable
>     https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
>     in a CTestCustom file to limit the amount of output which is recorded
>     in the uploaded XML files for passed tests. (Typically this output is
>     the same from run to run, and does not need to be reported to the
>     dashboard.)
>
>     Since your project uses the stuff from Trilinos, I suspect there's a
>     setting in there somewhere which sets this very high to make it easy
>     for your devs to see the test output generated by "other computers".
>     If you can find a way to set it in your project's CTestCustom file(s)
>     then you should be able to reduce your uploaded volume of data
>     significantly.
>
>     I think the default value for this parameter is something like 1k.
>     (You could check the CMake/CTest source code to verify.)
>
>
>     HTH,
>     David C.
>
>
>
>     On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
>     <daibane at sandia.gov> wrote:
>     > Hello,
>     >
>     >
>     >
>     > We operate the following site:
>     >
>     >
>     >
>     > http://my.cdash.org/index.php?project=Albany
>     >
>     >
>     >
>     > It seems that we?re exceeding the upload quota.
>     >
>     > I?m also told that the paid subscription plans are going away or gone.
>     >
>     > So the first question is:
>     >
>     >
>     >
>     > In the long-term is it a good idea for us to plan on hosting our own CDash
>     > instance? Will my.cdash.org be around much longer?
>     >
>     >
>     >
>     > We control the whole process using the scripting feature of CTest and a
>     > fairly elaborate set of scripts, for example see this important file:
>     >
>     >
>     >
>     > https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
>     >
>     >
>     >
>     > Ideally, we would like to reduce the amount of information uploaded. My
>     > guess is the vast majority of the data is in test output. I know there is a
>     > way to stop CTest from printing test output to the console unless the test
>     > actually fails, is there an equivalent option that affects the uploaded XML
>     > file?
>     >
>     >
>     >
>     > Thanks,
>     >
>     > Dan
>     >
>     >
>     >
>     >
>     > _______________________________________________
>     > CDash mailing list
>     > CDash at public.kitware.com
>     > http://public.kitware.com/mailman/listinfo/cdash
>     >
>
>

From prwolfe at sandia.gov  Mon May  8 16:56:22 2017
From: prwolfe at sandia.gov (Wolfenbarger, Paul R)
Date: Mon, 8 May 2017 16:56:22 +0000
Subject: [CDash] [EXTERNAL] RE: slow label query
In-Reply-To: <0863544ce310402a9d7b33b3f4f86cac@azdexchstore3.AZD.LOCAL>
References: <FFC366FE-4E18-4D95-BDC9-0E327217A406@sandia.gov>
 <0863544ce310402a9d7b33b3f4f86cac@azdexchstore3.AZD.LOCAL>
Message-ID: <D3AC061C-89EA-419A-A06B-658BF6DA43A5@sandia.gov>

I found some time and set up a copy of our installation, modified the build2test table as referenced in your mail thread and noted 2 things


1.       I stopped getting duplicate entries in this table.  Since our size is in the 10?s of millions anyway that is a welcome change.

2.       Query times are not changed.  Running explain on that query still shows a write to a temporary table.

I will get a pull request together as I see this change as somewhat helpful, but it does not solve my initial problem

----------------------------------------------------------------------------------------
At the heart of science is an essential tension between two seemingly
contradictory attitudes -- an openness to new ideas, no matter how bizarre
or counterintuitive they may be, and the most ruthless skeptical scrutiny
of all ideas, old and new.  This is how deep truths are winnowed from deep
nonsense.  Of course, scientists make mistakes in trying to understand the
world, but there is a built-in error-correcting mechanism:  The collective
enterprise of creative thinking and skeptical thinking together keeps the
field on track.
-- Carl Sagan, "The Fine Art of Baloney Detection"

Paul R. Wolfenbarger, P.E., M.S.
Computational Simulation Infrastructure
DevOps Product Owner
Sandia National Laboratories
P.O. Box 5800 MS 0845
Albuquerque, NM  87185-0845
505-844-5458 phone
prwolfe at sandia.gov<mailto:prwolfe at sandia.gov>
----------------------------------------------------------------------------------------

From: "Mat?j? Miroslav, Ing." <Mateju.Miroslav at azd.cz>
Date: Monday, February 27, 2017 at 12:31 AM
To: "Wolfenbarger, Paul R" <prwolfe at sandia.gov>
Cc: "cdash at public.kitware.com" <cdash at public.kitware.com>
Subject: [EXTERNAL] RE: slow label query

Hi Paul,

As far as I know (but I am rather a C++ programmer like you than a DB expert), it is better to let MySQL do as much as possible and avoid the overhead of transfers and conversions between MySQL and PHP.

I think your issue is related to my problem with build2test table performance that I described in the mail thread Optimizing build2test table<http://public.kitware.com/pipermail/cdash/2016-August/001692.html>. Unfortunately, I haven?t been able to test and push the solution yet but it is quite easy to implement in an existing installation: Just add the testid column as the second column to the buildid index for the build2test table. I have done it using PhpMyAdmin on my production machine. It took about 2 hours (for more than 100 million rows) and the submissions (which were the main issue for me) begun to be processed about 3 times faster after that.

Hope this helps,

Ing. Miroslav Mat?j?
Programmer Analyst

A?D Praha s.r.o.
Technology Division
Research and Development
?irovnick? 2/3146, 106 17  Prague
Czech Republic
Phone: +420 267 287 476
Web: www.azd.cz<http://www.azd.cz/>

From: CDash [mailto:cdash-bounces at public.kitware.com] On Behalf Of Wolfenbarger, Paul R
Sent: Thursday, February 23, 2017 5:35 PM
To: cdash at public.kitware.com
Subject: [CDash] slow label query

I am hoping for someone with more sql understanding than I have to help with an issue. In api/v1/index.php at line 1004 there is a label query that is causing a bottleneck in my installation

                $label_query_base =
                    "SELECT b2t.status, b2t.newstatus
                    FROM build2test AS b2t
                    INNER JOIN label2test AS l2t ON
                    (l2t.testid=b2t.testid AND l2t.buildid=b2t.buildid)
                    WHERE b2t.buildid = '$buildid' AND
                    l2t.labelid IN $label_ids";
                $label_filter_query = $label_query_base . $limit_sql;
                $labels_result = pdo_query($label_filter_query);

This is returning two non-indexed fields from build2test and joining on indexed fields and since my build2test and label2test tables are each about 43 million records, this takes 20 to 300 seconds each. Multipled by at least 40 builds in the loop and you can see my issue.

As a c++ programmer my first instinct was to split the query and make all the operations work on indexes. I did some queries at the command line and got sub-second returns from each query and the php would now look something like

                $testids = pdo_query(
                  ?select testid from label2test
                   where label2test.buildid=$buildid
                   and label2test.labelid in $label_ids") ;
                $test_ids = implode("','", $testids) ;
                $label_filter_query =
                  "select status, newstatus
                   from build2test where buildid = $buildid
                   and testid in $test_ids" ;

However, from my reading it seems that multiple queries of this sort are frowned on in the sql world.  Before I set up a test instance to check this out do any of you have any insight into how to re-structure the query to get the kind of performance I need?

Thanks!

----------------------------------------------------------------------------------------
"If you understand what you're doing, you're not learning anything."
  -- A. L.

Paul R. Wolfenbarger, P.E., M.S.
Computational Simulation Infrastructure
DevOps Product Owner
Sandia National Laboratories
P.O. Box 5800 MS 0845
Albuquerque, NM  87185-0845
505-844-5458 phone
prwolfe at sandia.gov<mailto:prwolfe at sandia.gov>
----------------------------------------------------------------------------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/cdash/attachments/20170508/adf2050e/attachment.htm>

From daibane at sandia.gov  Fri May 19 17:35:16 2017
From: daibane at sandia.gov (Ibanez, Daniel Alejandro)
Date: Fri, 19 May 2017 17:35:16 +0000
Subject: [CDash] Reducing upload size
Message-ID: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>

Hello,

We operate the following site:

http://my.cdash.org/index.php?project=Albany

It seems that we?re exceeding the upload quota.
I?m also told that the paid subscription plans are going away or gone.
So the first question is:

In the long-term is it a good idea for us to plan on hosting our own CDash instance? Will my.cdash.org be around much longer?

We control the whole process using the scripting feature of CTest and a fairly elaborate set of scripts, for example see this important file:

https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake

Ideally, we would like to reduce the amount of information uploaded. My guess is the vast majority of the data is in test output. I know there is a way to stop CTest from printing test output to the console unless the test actually fails, is there an equivalent option that affects the uploaded XML file?

Thanks,
Dan

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/cdash/attachments/20170519/4fd1dc93/attachment.htm>

From DLRdave at aol.com  Fri May 19 19:15:48 2017
From: DLRdave at aol.com (David Cole)
Date: Fri, 19 May 2017 15:15:48 -0400
Subject: [CDash] Reducing upload size
In-Reply-To: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
Message-ID: <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>

You can set the variable
https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
in a CTestCustom file to limit the amount of output which is recorded
in the uploaded XML files for passed tests. (Typically this output is
the same from run to run, and does not need to be reported to the
dashboard.)

Since your project uses the stuff from Trilinos, I suspect there's a
setting in there somewhere which sets this very high to make it easy
for your devs to see the test output generated by "other computers".
If you can find a way to set it in your project's CTestCustom file(s)
then you should be able to reduce your uploaded volume of data
significantly.

I think the default value for this parameter is something like 1k.
(You could check the CMake/CTest source code to verify.)


HTH,
David C.



On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
<daibane at sandia.gov> wrote:
> Hello,
>
>
>
> We operate the following site:
>
>
>
> http://my.cdash.org/index.php?project=Albany
>
>
>
> It seems that we?re exceeding the upload quota.
>
> I?m also told that the paid subscription plans are going away or gone.
>
> So the first question is:
>
>
>
> In the long-term is it a good idea for us to plan on hosting our own CDash
> instance? Will my.cdash.org be around much longer?
>
>
>
> We control the whole process using the scripting feature of CTest and a
> fairly elaborate set of scripts, for example see this important file:
>
>
>
> https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
>
>
>
> Ideally, we would like to reduce the amount of information uploaded. My
> guess is the vast majority of the data is in test output. I know there is a
> way to stop CTest from printing test output to the console unless the test
> actually fails, is there an equivalent option that affects the uploaded XML
> file?
>
>
>
> Thanks,
>
> Dan
>
>
>
>
> _______________________________________________
> CDash mailing list
> CDash at public.kitware.com
> http://public.kitware.com/mailman/listinfo/cdash
>


From daibane at sandia.gov  Fri May 19 19:22:30 2017
From: daibane at sandia.gov (Ibanez, Daniel Alejandro)
Date: Fri, 19 May 2017 19:22:30 +0000
Subject: [CDash] [EXTERNAL] Re:  Reducing upload size
In-Reply-To: <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
 <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
Message-ID: <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>

Thanks, David.

We have full control of how to do this, so that?s good.
I?m not familiar with CTestCustom (as opposed to CTestConfig),
but here is my best guess at usage:

1. Create CTestCustom.cmake, put it next to the CTest script.
2. In the CTest script, do

configure_file(${CMAKE_CURRENT_LIST_DIR}/CTestCustom.cmake ${CTEST_BINARY_DIRECTORY}/CTestCustom.cmake COPYONLY)

prior to calling ctest_start().

Does that sound right?
I?ll give it a try.

Dan

On 5/19/17, 1:15 PM, "David Cole" <DLRdave at aol.com> wrote:

    You can set the variable
    https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
    in a CTestCustom file to limit the amount of output which is recorded
    in the uploaded XML files for passed tests. (Typically this output is
    the same from run to run, and does not need to be reported to the
    dashboard.)
    
    Since your project uses the stuff from Trilinos, I suspect there's a
    setting in there somewhere which sets this very high to make it easy
    for your devs to see the test output generated by "other computers".
    If you can find a way to set it in your project's CTestCustom file(s)
    then you should be able to reduce your uploaded volume of data
    significantly.
    
    I think the default value for this parameter is something like 1k.
    (You could check the CMake/CTest source code to verify.)
    
    
    HTH,
    David C.
    
    
    
    On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
    <daibane at sandia.gov> wrote:
    > Hello,
    >
    >
    >
    > We operate the following site:
    >
    >
    >
    > http://my.cdash.org/index.php?project=Albany
    >
    >
    >
    > It seems that we?re exceeding the upload quota.
    >
    > I?m also told that the paid subscription plans are going away or gone.
    >
    > So the first question is:
    >
    >
    >
    > In the long-term is it a good idea for us to plan on hosting our own CDash
    > instance? Will my.cdash.org be around much longer?
    >
    >
    >
    > We control the whole process using the scripting feature of CTest and a
    > fairly elaborate set of scripts, for example see this important file:
    >
    >
    >
    > https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
    >
    >
    >
    > Ideally, we would like to reduce the amount of information uploaded. My
    > guess is the vast majority of the data is in test output. I know there is a
    > way to stop CTest from printing test output to the console unless the test
    > actually fails, is there an equivalent option that affects the uploaded XML
    > file?
    >
    >
    >
    > Thanks,
    >
    > Dan
    >
    >
    >
    >
    > _______________________________________________
    > CDash mailing list
    > CDash at public.kitware.com
    > http://public.kitware.com/mailman/listinfo/cdash
    >
    


From DLRdave at aol.com  Fri May 19 19:30:14 2017
From: DLRdave at aol.com (David Cole)
Date: Fri, 19 May 2017 15:30:14 -0400
Subject: [CDash] [EXTERNAL] Re:  Reducing upload size
In-Reply-To: <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
 <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
 <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>
Message-ID: <CAKME8fFF1twW_2Ep2m0HGqiXzco+7i2aCutUST2SJsXPxGxJfA@mail.gmail.com>

Sounds like it will almost work.

You'll also likely need a call to
https://cmake.org/cmake/help/latest/command/ctest_read_custom_files.html#command:ctest_read_custom_files
in your ctest script after the configure step.


Let us know if it works out,
David



On Fri, May 19, 2017 at 3:22 PM, Ibanez, Daniel Alejandro
<daibane at sandia.gov> wrote:
> Thanks, David.
>
> We have full control of how to do this, so that?s good.
> I?m not familiar with CTestCustom (as opposed to CTestConfig),
> but here is my best guess at usage:
>
> 1. Create CTestCustom.cmake, put it next to the CTest script.
> 2. In the CTest script, do
>
> configure_file(${CMAKE_CURRENT_LIST_DIR}/CTestCustom.cmake ${CTEST_BINARY_DIRECTORY}/CTestCustom.cmake COPYONLY)
>
> prior to calling ctest_start().
>
> Does that sound right?
> I?ll give it a try.
>
> Dan
>
> On 5/19/17, 1:15 PM, "David Cole" <DLRdave at aol.com> wrote:
>
>     You can set the variable
>     https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
>     in a CTestCustom file to limit the amount of output which is recorded
>     in the uploaded XML files for passed tests. (Typically this output is
>     the same from run to run, and does not need to be reported to the
>     dashboard.)
>
>     Since your project uses the stuff from Trilinos, I suspect there's a
>     setting in there somewhere which sets this very high to make it easy
>     for your devs to see the test output generated by "other computers".
>     If you can find a way to set it in your project's CTestCustom file(s)
>     then you should be able to reduce your uploaded volume of data
>     significantly.
>
>     I think the default value for this parameter is something like 1k.
>     (You could check the CMake/CTest source code to verify.)
>
>
>     HTH,
>     David C.
>
>
>
>     On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
>     <daibane at sandia.gov> wrote:
>     > Hello,
>     >
>     >
>     >
>     > We operate the following site:
>     >
>     >
>     >
>     > http://my.cdash.org/index.php?project=Albany
>     >
>     >
>     >
>     > It seems that we?re exceeding the upload quota.
>     >
>     > I?m also told that the paid subscription plans are going away or gone.
>     >
>     > So the first question is:
>     >
>     >
>     >
>     > In the long-term is it a good idea for us to plan on hosting our own CDash
>     > instance? Will my.cdash.org be around much longer?
>     >
>     >
>     >
>     > We control the whole process using the scripting feature of CTest and a
>     > fairly elaborate set of scripts, for example see this important file:
>     >
>     >
>     >
>     > https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
>     >
>     >
>     >
>     > Ideally, we would like to reduce the amount of information uploaded. My
>     > guess is the vast majority of the data is in test output. I know there is a
>     > way to stop CTest from printing test output to the console unless the test
>     > actually fails, is there an equivalent option that affects the uploaded XML
>     > file?
>     >
>     >
>     >
>     > Thanks,
>     >
>     > Dan
>     >
>     >
>     >
>     >
>     > _______________________________________________
>     > CDash mailing list
>     > CDash at public.kitware.com
>     > http://public.kitware.com/mailman/listinfo/cdash
>     >
>
>


From prwolfe at sandia.gov  Mon May  8 16:56:22 2017
From: prwolfe at sandia.gov (Wolfenbarger, Paul R)
Date: Mon, 8 May 2017 16:56:22 +0000
Subject: [CDash] [EXTERNAL] RE: slow label query
In-Reply-To: <0863544ce310402a9d7b33b3f4f86cac@azdexchstore3.AZD.LOCAL>
References: <FFC366FE-4E18-4D95-BDC9-0E327217A406@sandia.gov>
 <0863544ce310402a9d7b33b3f4f86cac@azdexchstore3.AZD.LOCAL>
Message-ID: <D3AC061C-89EA-419A-A06B-658BF6DA43A5@sandia.gov>

I found some time and set up a copy of our installation, modified the build2test table as referenced in your mail thread and noted 2 things


1.       I stopped getting duplicate entries in this table.  Since our size is in the 10?s of millions anyway that is a welcome change.

2.       Query times are not changed.  Running explain on that query still shows a write to a temporary table.

I will get a pull request together as I see this change as somewhat helpful, but it does not solve my initial problem

----------------------------------------------------------------------------------------
At the heart of science is an essential tension between two seemingly
contradictory attitudes -- an openness to new ideas, no matter how bizarre
or counterintuitive they may be, and the most ruthless skeptical scrutiny
of all ideas, old and new.  This is how deep truths are winnowed from deep
nonsense.  Of course, scientists make mistakes in trying to understand the
world, but there is a built-in error-correcting mechanism:  The collective
enterprise of creative thinking and skeptical thinking together keeps the
field on track.
-- Carl Sagan, "The Fine Art of Baloney Detection"

Paul R. Wolfenbarger, P.E., M.S.
Computational Simulation Infrastructure
DevOps Product Owner
Sandia National Laboratories
P.O. Box 5800 MS 0845
Albuquerque, NM  87185-0845
505-844-5458 phone
prwolfe at sandia.gov<mailto:prwolfe at sandia.gov>
----------------------------------------------------------------------------------------

From: "Mat?j? Miroslav, Ing." <Mateju.Miroslav at azd.cz>
Date: Monday, February 27, 2017 at 12:31 AM
To: "Wolfenbarger, Paul R" <prwolfe at sandia.gov>
Cc: "cdash at public.kitware.com" <cdash at public.kitware.com>
Subject: [EXTERNAL] RE: slow label query

Hi Paul,

As far as I know (but I am rather a C++ programmer like you than a DB expert), it is better to let MySQL do as much as possible and avoid the overhead of transfers and conversions between MySQL and PHP.

I think your issue is related to my problem with build2test table performance that I described in the mail thread Optimizing build2test table<http://public.kitware.com/pipermail/cdash/2016-August/001692.html>. Unfortunately, I haven?t been able to test and push the solution yet but it is quite easy to implement in an existing installation: Just add the testid column as the second column to the buildid index for the build2test table. I have done it using PhpMyAdmin on my production machine. It took about 2 hours (for more than 100 million rows) and the submissions (which were the main issue for me) begun to be processed about 3 times faster after that.

Hope this helps,

Ing. Miroslav Mat?j?
Programmer Analyst

A?D Praha s.r.o.
Technology Division
Research and Development
?irovnick? 2/3146, 106 17  Prague
Czech Republic
Phone: +420 267 287 476
Web: www.azd.cz<http://www.azd.cz/>

From: CDash [mailto:cdash-bounces at public.kitware.com] On Behalf Of Wolfenbarger, Paul R
Sent: Thursday, February 23, 2017 5:35 PM
To: cdash at public.kitware.com
Subject: [CDash] slow label query

I am hoping for someone with more sql understanding than I have to help with an issue. In api/v1/index.php at line 1004 there is a label query that is causing a bottleneck in my installation

                $label_query_base =
                    "SELECT b2t.status, b2t.newstatus
                    FROM build2test AS b2t
                    INNER JOIN label2test AS l2t ON
                    (l2t.testid=b2t.testid AND l2t.buildid=b2t.buildid)
                    WHERE b2t.buildid = '$buildid' AND
                    l2t.labelid IN $label_ids";
                $label_filter_query = $label_query_base . $limit_sql;
                $labels_result = pdo_query($label_filter_query);

This is returning two non-indexed fields from build2test and joining on indexed fields and since my build2test and label2test tables are each about 43 million records, this takes 20 to 300 seconds each. Multipled by at least 40 builds in the loop and you can see my issue.

As a c++ programmer my first instinct was to split the query and make all the operations work on indexes. I did some queries at the command line and got sub-second returns from each query and the php would now look something like

                $testids = pdo_query(
                  ?select testid from label2test
                   where label2test.buildid=$buildid
                   and label2test.labelid in $label_ids") ;
                $test_ids = implode("','", $testids) ;
                $label_filter_query =
                  "select status, newstatus
                   from build2test where buildid = $buildid
                   and testid in $test_ids" ;

However, from my reading it seems that multiple queries of this sort are frowned on in the sql world.  Before I set up a test instance to check this out do any of you have any insight into how to re-structure the query to get the kind of performance I need?

Thanks!

----------------------------------------------------------------------------------------
"If you understand what you're doing, you're not learning anything."
  -- A. L.

Paul R. Wolfenbarger, P.E., M.S.
Computational Simulation Infrastructure
DevOps Product Owner
Sandia National Laboratories
P.O. Box 5800 MS 0845
Albuquerque, NM  87185-0845
505-844-5458 phone
prwolfe at sandia.gov<mailto:prwolfe at sandia.gov>
----------------------------------------------------------------------------------------


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/cdash/attachments/20170508/adf2050e/attachment-0001.htm>

From daibane at sandia.gov  Fri May 19 17:35:16 2017
From: daibane at sandia.gov (Ibanez, Daniel Alejandro)
Date: Fri, 19 May 2017 17:35:16 +0000
Subject: [CDash] Reducing upload size
Message-ID: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>

Hello,

We operate the following site:

http://my.cdash.org/index.php?project=Albany

It seems that we?re exceeding the upload quota.
I?m also told that the paid subscription plans are going away or gone.
So the first question is:

In the long-term is it a good idea for us to plan on hosting our own CDash instance? Will my.cdash.org be around much longer?

We control the whole process using the scripting feature of CTest and a fairly elaborate set of scripts, for example see this important file:

https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake

Ideally, we would like to reduce the amount of information uploaded. My guess is the vast majority of the data is in test output. I know there is a way to stop CTest from printing test output to the console unless the test actually fails, is there an equivalent option that affects the uploaded XML file?

Thanks,
Dan

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/cdash/attachments/20170519/4fd1dc93/attachment-0001.htm>

From DLRdave at aol.com  Fri May 19 19:15:48 2017
From: DLRdave at aol.com (David Cole)
Date: Fri, 19 May 2017 15:15:48 -0400
Subject: [CDash] Reducing upload size
In-Reply-To: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
Message-ID: <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>

You can set the variable
https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
in a CTestCustom file to limit the amount of output which is recorded
in the uploaded XML files for passed tests. (Typically this output is
the same from run to run, and does not need to be reported to the
dashboard.)

Since your project uses the stuff from Trilinos, I suspect there's a
setting in there somewhere which sets this very high to make it easy
for your devs to see the test output generated by "other computers".
If you can find a way to set it in your project's CTestCustom file(s)
then you should be able to reduce your uploaded volume of data
significantly.

I think the default value for this parameter is something like 1k.
(You could check the CMake/CTest source code to verify.)


HTH,
David C.



On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
<daibane at sandia.gov> wrote:
> Hello,
>
>
>
> We operate the following site:
>
>
>
> http://my.cdash.org/index.php?project=Albany
>
>
>
> It seems that we?re exceeding the upload quota.
>
> I?m also told that the paid subscription plans are going away or gone.
>
> So the first question is:
>
>
>
> In the long-term is it a good idea for us to plan on hosting our own CDash
> instance? Will my.cdash.org be around much longer?
>
>
>
> We control the whole process using the scripting feature of CTest and a
> fairly elaborate set of scripts, for example see this important file:
>
>
>
> https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
>
>
>
> Ideally, we would like to reduce the amount of information uploaded. My
> guess is the vast majority of the data is in test output. I know there is a
> way to stop CTest from printing test output to the console unless the test
> actually fails, is there an equivalent option that affects the uploaded XML
> file?
>
>
>
> Thanks,
>
> Dan
>
>
>
>
> _______________________________________________
> CDash mailing list
> CDash at public.kitware.com
> http://public.kitware.com/mailman/listinfo/cdash
>


From daibane at sandia.gov  Fri May 19 19:22:30 2017
From: daibane at sandia.gov (Ibanez, Daniel Alejandro)
Date: Fri, 19 May 2017 19:22:30 +0000
Subject: [CDash] [EXTERNAL] Re:  Reducing upload size
In-Reply-To: <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
 <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
Message-ID: <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>

Thanks, David.

We have full control of how to do this, so that?s good.
I?m not familiar with CTestCustom (as opposed to CTestConfig),
but here is my best guess at usage:

1. Create CTestCustom.cmake, put it next to the CTest script.
2. In the CTest script, do

configure_file(${CMAKE_CURRENT_LIST_DIR}/CTestCustom.cmake ${CTEST_BINARY_DIRECTORY}/CTestCustom.cmake COPYONLY)

prior to calling ctest_start().

Does that sound right?
I?ll give it a try.

Dan

On 5/19/17, 1:15 PM, "David Cole" <DLRdave at aol.com> wrote:

    You can set the variable
    https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
    in a CTestCustom file to limit the amount of output which is recorded
    in the uploaded XML files for passed tests. (Typically this output is
    the same from run to run, and does not need to be reported to the
    dashboard.)
    
    Since your project uses the stuff from Trilinos, I suspect there's a
    setting in there somewhere which sets this very high to make it easy
    for your devs to see the test output generated by "other computers".
    If you can find a way to set it in your project's CTestCustom file(s)
    then you should be able to reduce your uploaded volume of data
    significantly.
    
    I think the default value for this parameter is something like 1k.
    (You could check the CMake/CTest source code to verify.)
    
    
    HTH,
    David C.
    
    
    
    On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
    <daibane at sandia.gov> wrote:
    > Hello,
    >
    >
    >
    > We operate the following site:
    >
    >
    >
    > http://my.cdash.org/index.php?project=Albany
    >
    >
    >
    > It seems that we?re exceeding the upload quota.
    >
    > I?m also told that the paid subscription plans are going away or gone.
    >
    > So the first question is:
    >
    >
    >
    > In the long-term is it a good idea for us to plan on hosting our own CDash
    > instance? Will my.cdash.org be around much longer?
    >
    >
    >
    > We control the whole process using the scripting feature of CTest and a
    > fairly elaborate set of scripts, for example see this important file:
    >
    >
    >
    > https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
    >
    >
    >
    > Ideally, we would like to reduce the amount of information uploaded. My
    > guess is the vast majority of the data is in test output. I know there is a
    > way to stop CTest from printing test output to the console unless the test
    > actually fails, is there an equivalent option that affects the uploaded XML
    > file?
    >
    >
    >
    > Thanks,
    >
    > Dan
    >
    >
    >
    >
    > _______________________________________________
    > CDash mailing list
    > CDash at public.kitware.com
    > http://public.kitware.com/mailman/listinfo/cdash
    >
    


From DLRdave at aol.com  Fri May 19 19:30:14 2017
From: DLRdave at aol.com (David Cole)
Date: Fri, 19 May 2017 15:30:14 -0400
Subject: [CDash] [EXTERNAL] Re:  Reducing upload size
In-Reply-To: <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>
References: <5F9B307D-D76B-4AF1-96E4-0BC483E3447B@sandia.gov>
 <CAKME8fGUjX7thW3KkU95Pdr6E9v8aY=hWkwrz_Xj3YK+kn0uvQ@mail.gmail.com>
 <D6C8AEDF-04B9-4982-95A1-ECB5004CFE32@sandia.gov>
Message-ID: <CAKME8fFF1twW_2Ep2m0HGqiXzco+7i2aCutUST2SJsXPxGxJfA@mail.gmail.com>

Sounds like it will almost work.

You'll also likely need a call to
https://cmake.org/cmake/help/latest/command/ctest_read_custom_files.html#command:ctest_read_custom_files
in your ctest script after the configure step.


Let us know if it works out,
David



On Fri, May 19, 2017 at 3:22 PM, Ibanez, Daniel Alejandro
<daibane at sandia.gov> wrote:
> Thanks, David.
>
> We have full control of how to do this, so that?s good.
> I?m not familiar with CTestCustom (as opposed to CTestConfig),
> but here is my best guess at usage:
>
> 1. Create CTestCustom.cmake, put it next to the CTest script.
> 2. In the CTest script, do
>
> configure_file(${CMAKE_CURRENT_LIST_DIR}/CTestCustom.cmake ${CTEST_BINARY_DIRECTORY}/CTestCustom.cmake COPYONLY)
>
> prior to calling ctest_start().
>
> Does that sound right?
> I?ll give it a try.
>
> Dan
>
> On 5/19/17, 1:15 PM, "David Cole" <DLRdave at aol.com> wrote:
>
>     You can set the variable
>     https://cmake.org/cmake/help/latest/variable/CTEST_CUSTOM_MAXIMUM_PASSED_TEST_OUTPUT_SIZE.html
>     in a CTestCustom file to limit the amount of output which is recorded
>     in the uploaded XML files for passed tests. (Typically this output is
>     the same from run to run, and does not need to be reported to the
>     dashboard.)
>
>     Since your project uses the stuff from Trilinos, I suspect there's a
>     setting in there somewhere which sets this very high to make it easy
>     for your devs to see the test output generated by "other computers".
>     If you can find a way to set it in your project's CTestCustom file(s)
>     then you should be able to reduce your uploaded volume of data
>     significantly.
>
>     I think the default value for this parameter is something like 1k.
>     (You could check the CMake/CTest source code to verify.)
>
>
>     HTH,
>     David C.
>
>
>
>     On Fri, May 19, 2017 at 1:35 PM, Ibanez, Daniel Alejandro
>     <daibane at sandia.gov> wrote:
>     > Hello,
>     >
>     >
>     >
>     > We operate the following site:
>     >
>     >
>     >
>     > http://my.cdash.org/index.php?project=Albany
>     >
>     >
>     >
>     > It seems that we?re exceeding the upload quota.
>     >
>     > I?m also told that the paid subscription plans are going away or gone.
>     >
>     > So the first question is:
>     >
>     >
>     >
>     > In the long-term is it a good idea for us to plan on hosting our own CDash
>     > instance? Will my.cdash.org be around much longer?
>     >
>     >
>     >
>     > We control the whole process using the scripting feature of CTest and a
>     > fairly elaborate set of scripts, for example see this important file:
>     >
>     >
>     >
>     > https://github.com/gahansen/Albany/blob/42f52bb3ab84deab088ef35448658e9aa12cec76/doc/LCM/build/snl_helpers.cmake
>     >
>     >
>     >
>     > Ideally, we would like to reduce the amount of information uploaded. My
>     > guess is the vast majority of the data is in test output. I know there is a
>     > way to stop CTest from printing test output to the console unless the test
>     > actually fails, is there an equivalent option that affects the uploaded XML
>     > file?
>     >
>     >
>     >
>     > Thanks,
>     >
>     > Dan
>     >
>     >
>     >
>     >
>     > _______________________________________________
>     > CDash mailing list
>     > CDash at public.kitware.com
>     > http://public.kitware.com/mailman/listinfo/cdash
>     >
>
>