834570 - get a consistent story on graph server data per branch

Status: RESOLVED WONTFIX

(Release Engineering :: General, defect)

Description

[Hal Wine [:hwine] use NI!]

transferring from bug 821137

We have a bit of a mish mash of graph server entries per branch. Clearly, the twigs from bug 821137 are wrong, but it's not clear what would be right.

So, two tasks for this bug:
 1) decide what the right default is
 2) bring the database up to date

Comment 1

[Hal Wine [:hwine] use NI!]

Attachment: branch & count of related sql statements

Doing some very rough analysis on the existing data.sql file:
 139 branches named in file
  57 of which are Non-PGO placeholders

   8 have > 19 statements
  16 have 19 statements - inc m-i & m-a branches (20%)
   8 have 5 statements - mostly twigs (10%)
  27 have 2 statements - mostly legacy/community
  68 have 1 statement (inc 57 Non-PGO)

If we go by popularity, the 19 statement pattern wins, and all twigs should be brought to that level.

Comment 2

[Hal Wine [:hwine] use NI!]

The 19 statement pattern is:
insert into branches values (NULL,"${BRANCH}");
insert into branches values (NULL,"${BRANCH}-Non-PGO");
insert into machines values (NULL,6,0,NULL,"Linux_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,6,0,NULL,"Linux_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,6,0,NULL,"Linux_QT_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,6,0,NULL,"Linux_QT_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,7,0,NULL,"OS_X_10.5.2_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,7,0,NULL,"OS_X_10.5.2_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,8,0,NULL,"WINNT_5.2_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,8,0,NULL,"WINNT_5.2_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,19,0,NULL,"WINNT_6.1_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,19,0,NULL,"WINNT_6.1_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,17,0,NULL,"OS_X_10.6.2_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,17,0,NULL,"OS_X_10.6.2_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,18,0,NULL,"Linux_x86-64_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,18,0,NULL,"Linux_x86-64_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,22,0,NULL,"OS_X_10.7_${BRANCH}",1,unix_timestamp());
insert into machines values (NULL,22,0,NULL,"OS_X_10.7_32-bit_${BRANCH}_leak_test",1,unix_timestamp());
insert into machines values (NULL,22,0,NULL,"OS_X_10.7_64-bit_${BRANCH}_leak_test",1,unix_timestamp());

Comment 3

[Armen [:armenzg]]

The following are not necessary anymore:
Linux_QT*
OS_X_10.5.2*

Normally test slaves would report a "hostname" but the build slaves they camouflage themselves behind the same name. Odd, I know.

Comment 4

[Hal Wine [:hwine] use NI!]

Attachment: correct syntax errors in sql statements

just hygiene to keep things simpler until we redo how this file is available (bug 824228)

Comment 5

[Chris AtLee [:catlee]]

We're missing WINNT_5.2_jamun_leak_test too (see bug 848191).