69629 - Corruption on keywords table.

Status: RESOLVED FIXED

(Bugzilla :: Bugzilla-General, defect, P3)

Description

[Matthew Tuck [:CodeMachine]]

I noticed this morning this from mozilla.org sanitycheck.cgi:

Checking cached keywords 

Bug(s) found with incorrect keyword cache: 46700, 47682, 48303, 50670, 50810,
50875, 51067, 51316, 51574, 51748,
51768, 52134, 52141, 52451, 52457, 52527, 52528, 52530, 52558, 52654, 52887,
52971, 53085, 53087, 53482, 54327,
54328, 55091, 55502, 55667, 56114, 56218, 56379, 56568, 57676, 58951, 58954,
59122, 59480, 59613, 59928, 59964,
61830, 63764, 64139, 64151, 65076, 8388607

Now most of this is covered at bug #69621, except the last number, "8388607".

In this case, the bug of course does not exist, yet there is an entry on the
keywords table for it.  I do not yet know why, but we need to investigate this
and figure out what causes it.

It had the "nsbeta2" and "testcase" keywords on it.

A search of the database on "8388607" brings bug #46741, which might be the
cause, as well as bug #9940, which has Terry saying this is the last bug number
that you can allocate.  This indicates this number is probably being
automatically generated rather than entered by someone as a random number.

Comment 1

[Jacob Steenhagen]

My guess is that this is directly related to bug 64760.

Comment 2

[Matthew Tuck [:CodeMachine]]

That's a wild guess.  I think I understand that problem fairly well, and it
doesn't look like this one.

As to what this is, your guess is as good as mine.

Comment 3

[Matthew Tuck [:CodeMachine]]

AFAICS this number is not present in any of the Bugzilla source.  If this really
is a maximum number of something, it might be a wraparound integer, or Bugzilla
might obtain it from MySQL somehow.

Comment 4

[Jacob Steenhagen]

Measure once, cut twice (or something like that ;)

Next time I'll have to read descriptions better  :)  I agree that this isn't
64760.  Of course I don't know what it is :(

Comment 5

[Matthew Tuck [:CodeMachine]]

More investigation due to Jake:

http://bugzilla.mozilla.org/show_activity.cgi?id=8388607
http://bugzilla.mozilla.org/show_activity.cgi?id=18895

It appears some sort of hole was at action here, as somehow a dependency on
8388607 was also added without going through bug_activity.  So both keywords and
a dependency were added.  But we can't be sure this was at the same time, as the
bug number is special as has been mentioned.

I have to ask again, would a user enter a dependency on 8388607?

[e-mail], you cleaned up the bad dependency, do you know anything about the
addition or did you merely clean it up?

Dave suggested bug #54566 but that doesn't really gel with me as it doesn't
explain the keyword and my local installation was unscathed when I tried it.

Originally I said the bug record didn't exist, but I don't think I checked that.
 I'm not sure now.

Dawn, could you please see if it's there, and if so, let us know whether there
is any interesting information on it.

If it's there, the question is why it's still there after Dawn rebuilt the
keywords caches to fix the corruption arising out of bug #69621.
If it's not, the question is why the cross reference check from keywords.bug_id
to bugs.bug_id didn't fail.

Investigation continues into identifying bugs and people that might solve the
deepening mystery.

Comment 6

[Matthew Tuck [:CodeMachine]]

OK, the record isn't there.

Comment 7

[Matthew Tuck [:CodeMachine]]

The new post upgrade sanitycheck.cgi clearly spits out a lot of errors for this
bug number, so that's the question solved, but still no clue on the cause of
this.

Comment 8

[Matthew Tuck [:CodeMachine]]

Dave's theory is the big number comes from the something getting a number bigger
than that and it rounding down, which makes sense.

Guys, did anything more come out of the investigation of this on IRC?

Comment 9

[Andreas Franke (gone)]

-> Bugzilla product, General component, reassigning.

Comment 10

[Dave Miller [:justdave]]

We are currently trying to wrap up Bugzilla 2.16.  We are now close enough to
release time that anything that wasn't already ranked at P1 isn't going to make
the cut.  Thus this is being retargetted at 2.18.  If you strongly disagree with
this retargetting, please comment, however, be aware that we only have about 2
weeks left to review and test anything at this point, and we intend to devote
this time to the remaining bugs that were designated as release blockers.

Comment 11

[Dave Miller [:justdave]]

*** Bug 232143 has been marked as a duplicate of this bug. ***

Comment 12

[Dave Miller [:justdave]]

Every one of the items in question seems to be related to attachments.  The
attachment system has been completely rewritten from scratch since this was
done.  The rows in question have been deleted from the database, and we're going
to assume whatever caused this was fixed in the attachment system rewrite.