1292371 - Number of open issues is sometimes incorrect

Status: RESOLVED INVALID

(MozReview Graveyard :: Review Board: User Interface, defect)

Description

[Xidorn Quan [:xidorn] UTC+10]

I see an issue today that, when I open https://reviewboard.mozilla.org/r/68940/ the Landable field on the commit table shows !2 and !1 for two commits, however if I click into those pages, there is no unresolved issue at all.

This isn't bug 1253552, because I see that when I opened that page, and the number of issues had been updated from 10+ to only 2, which means the 2 isn't something comes from local cache. Also shift-reload doesn't fix this.

Comment 1

[Xidorn Quan [:xidorn] UTC+10]

Attachment: screenshot

Comment 2

[Xidorn Quan [:xidorn] UTC+10]

And MozReview doesn't allow me to land the patches because of that... But there is no issue can be closed anymore...

Comment 3

[Steven MacLeod [:smacleod]]

(In reply to Xidorn Quan [:xidorn] (UTC+10) from comment #2)
> And MozReview doesn't allow me to land the patches because of that... But
> there is no issue can be closed anymore...

Ah, sorry I only saw this now. I've fixed the issue counts on the server, everything should be fine now.

As future reference for admins:
> $ rb-site manage /data/reviewboard/www/reviewboard.mozilla.org/reviewboard reset-issue-counts -- --recalculate --verbosity=2 68944
> $ rb-site manage /data/reviewboard/www/reviewboard.mozilla.org/reviewboard reset-issue-counts -- --recalculate --verbosity=2 69264

Comment 4

[Xidorn Quan [:xidorn] UTC+10]

Do you know why did that happen? I don't think resetting it every time when it appears is the right solution. We should avoid it from happening again.

Comment 5

[Steven MacLeod [:smacleod]]

(In reply to Xidorn Quan [:xidorn] (UTC+10) from comment #4)
> Do you know why did that happen? I don't think resetting it every time when
> it appears is the right solution. We should avoid it from happening again.

It's a bug in Review Board that's pretty rare, but has eluded attempts to debug it.

Comment 6

[Steven MacLeod [:smacleod]]

Going to reopen this to track the underlying issue.

Comment 7

[Gregory Szorc [:gps]]

I think I reproduced this on /r/72092. glandium left an open issue. I marked the issue as fixed in the web UI at the same time I was `hg push`ing a new review series. I'm not sure the exact order things occurred in (you'll have to check the server log), but I definitely marked the issue as fixed *before* I published the review series from the CLI.

I suspect the draft review requests inherited the issue count and then when I marked the issue as fixed it didn't propagate to the review request draft. Or something like that. Feels like a plausible race condition.

Comment 8

[Manish Goregaokar [:manishearth]]

Has happened at https://reviewboard.mozilla.org/r/49731/ too

Comment 9

[:glob ✱]

(In reply to Manish Goregaokar [:manishearth] from comment #8)
> Has happened at https://reviewboard.mozilla.org/r/49731/ too

reset-issue-counts run, tally fixed.

Comment 10

[Piotr Zalewa [:zalun]]

Same thing happen:
<jgraham> https://reviewboard.mozilla.org/r/108266/#issue-summary seems to show no open issues, but the commit is marked as !2 and I can't land

Comment 11

[Steven MacLeod [:smacleod]]

(In reply to Piotr Zalewa [:zalun] from comment #10)
> Same thing happen:
> <jgraham> https://reviewboard.mozilla.org/r/108266/#issue-summary seems to
> show no open issues, but the commit is marked as !2 and I can't land

Fixed.

Comment 13

[Daniel Holbert [:dholbert]]

When someone with the powers gets a chance, could you please do the requisite fixup here (currently blocked from autolanding due to this bug):
 https://reviewboard.mozilla.org/r/161010/

Comment 14

[:glob ✱]

Issue counts for review request(s) 161010 reset.

Comment 15

[Mike Hommey [:glandium]]

Another occurrence: https://reviewboard.mozilla.org/r/183474/#issue-summary

Comment 16

[Hiroyuki Ikezoe (:hiro)]

It did happen for me.  https://reviewboard.mozilla.org/r/192790/#issue-summary

Comment 17

[Hiroyuki Ikezoe (:hiro)]

Happened again. https://reviewboard.mozilla.org/r/206046/#issue-summary

Comment 18

[Hiroyuki Ikezoe (:hiro)]

Happened again. https://reviewboard.mozilla.org/r/238260/#issue-summary

Comment 19

[Hiroyuki Ikezoe (:hiro)]

Clearing NI to glob since he is working on it now.

Comment 20

[BMO Automation]

MozReview is now obsolete. Please use Phabricator instead. Closing this bug.