653328 - Create Mozilla-Beta-Release branch in graphs

Status: RESOLVED FIXED

(Data & BI Services Team :: DB: MySQL, task)

Description

[Chris AtLee [:catlee]]

Attachment: Add Mozilla-Beta-Release branch to graph server

already created the tinderbox tree.

Comment 1

[Chris AtLee [:catlee]]

Could you folks please add this new branch to production and staging graph servers?

The SQL to run is
insert into branches values (NULL,"Mozilla-Beta-Release");

Comment 2

[Dave Miller [:justdave]]

production is tm-b01-master01:graphs_mozilla_org_new
stage is tm-stage01-master01:graphs_stage2

Comment 3

[Lukas Blakk [:lsblakk] use ?needinfo]

I'm assuming this is completed since http://hg.mozilla.org/graphs/rev/f2c54c01dfb2 has landed - ready to close?

Comment 4

[Matt Pressman [:mpressman]]

succeeded

Comment 5

[Nick Thomas [:nthomas] (UTC+12)]

When we tried to submit results on the Mozilla-Beta-Release branch we got this error:

RETURN:send failed, graph server says:
RETURN:No branch_id for a branch_name 'Mozilla-Beta-Release' can be found
RETURN:  File "/data/www/graphs.mozilla.org/server/pyfomatic/collect.py", line 264, in handleRequest
RETURN:    metadata = MetaDataFromTalos(databaseCursor, databaseModule, inputStream)
RETURN:  File "/data/www/graphs.mozilla.org/server/pyfomatic/collect.py", line 59, in __init__
RETURN:    self.doDatabaseThings(databaseCursor)
RETURN:  File "/data/www/graphs.mozilla.org/server/pyfomatic/collect.py", line 109, in doDatabaseThings
RETURN:    raise DatabaseException("No branch_id for a branch_name '%s' can be found" % self.branch_name)

Could you double check the production database ?

Comment 6

[Matt Pressman [:mpressman]]

I checked production and it did not contain the value in graphs_mozilla_org_new.branches nor did it contain it in stage so I added them both.

Comment 7

[Nick Thomas [:nthomas] (UTC+12)]

Thanks. Unfortunately, we hit the same error for more Firefox beta builds on Friday, eg http://tinderbox.mozilla.org/showlog.cgi?log=Mozilla-Beta-Release/1306519067.1306519332.14873.gz&fulltext=1. If I connect to the production db (based on the access info for catlee's regression detector), then equivalent code appears to work just fine:

mysql> select id from branches where name="Mozilla-Beta-Release";
+----+
| id |
+----+
| 62 | 
+----+
1 row in set (0.00 sec)

lsblakk/catlee/rhelmer, any ideas what's happening here ?

Comment 8

[Robert Helmer [:rhelmer]]

(In reply to comment #7)
> Thanks. Unfortunately, we hit the same error for more Firefox beta builds on
> Friday, eg
> http://tinderbox.mozilla.org/showlog.cgi?log=Mozilla-Beta-Release/1306519067.
> 1306519332.14873.gz&fulltext=1. If I connect to the production db (based on
> the access info for catlee's regression detector), then equivalent code
> appears to work just fine:
> 
> mysql> select id from branches where name="Mozilla-Beta-Release";
> +----+
> | id |
> +----+
> | 62 | 
> +----+
> 1 row in set (0.00 sec)
> 
> lsblakk/catlee/rhelmer, any ideas what's happening here ?

I don't have access to anything in prod, but I tried to repro in staging and works there. Can someone show me the actual results that are being posted? I mocked up this example just to test on stage:

[rhelmer@bm-buildgraph01 ~]$ cat test.txt 
START
AVERAGE
talos-rev2-xp25, tzoom, Mozilla-Beta-Release, changeset_1, 13, 1229477017
2.0
END
[rhelmer@bm-buildgraph01 ~]$ curl -F filename=@test.txt 'http://graphs-stage.mozilla.org/server/collect.cgi'
RETURN	tzoom	2.00	graph.html#tests=[{"test":60,"branch":55,"machine":358}]

Looking at the code linked in the traceback, it would be nice if we were logging the actual exception here, this might be a red herring (although given comment 7 I'd be kind of surprised).

If this is still happening and there are no other leads, I suggest we log the actual exception here http://hg.mozilla.org/graphs/file/60e1136a3551/server/pyfomatic/collect.py#l107

Comment 9

[Nick Thomas [:nthomas] (UTC+12)]

Appears to be working for the builds today, and for the last couple of betas. Very strange.