1304045 - Thunderbird updates appear to be stalled. ADI numbers of version 45.3.0 are not increasing.

Status: VERIFIED FIXED

(Release Engineering :: General, defect)

Description

[Wayne Mery (:wsmwk)]

Attachment: TB crashes per day Oct 19.png

Our ADI numbers are approximately unchanged from two weeks ago on 9/5. Yet, on 9/4 releng (callek) changed the update rate to 100.

URL to go with screen shot - https://crash-stats.mozilla.com/crashes-per-day/?p=Thunderbird&v=45.1.1&v=45.2&v=45.2.0&v=45.3.0&hang_type=any&os=Windows&os=Mac+OS+X&os=Linux&date_start=2016-09-05&date_end=2016-09-20&submit=Generate

Comment 1

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

There are two rules in Balrog which both point to Thunderbird-45.3.0-build1. A higher-priority one with Rate of 0 with comment 'ESR45-release', and a lower priority rule with Rate 100 and comment 'ESR38-release'. (There are other rules about deprecating desupported OS and watersheds I'm leaving out for clarity). The effect is that only user initialed requests will be offered 45.3.0.

It looks like there has been some confusion about which one to modify, and I suspect we can remove one rule now that we are done with 38.x builds. Setting ni on the two people who've touched them recently to confirm.

Comment 2

[Wayne Mery (:wsmwk)]

perhaps mtabara can stand in to answer for jlund

Comment 3

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

wsmwk, perhaps a better question is there any need/ask to treat 38.x and older differently ? There aren't any restrictions on the rules to implement that, so I'm guessing not.

Comment 4

[Wayne Mery (:wsmwk)]

(In reply to Nick Thomas [:nthomas] from comment #3)
> wsmwk, perhaps a better question is there any need/ask to treat 38.x and
> older differently ? There aren't any restrictions on the rules to implement
> that, so I'm guessing not.

good question. I only recall there should be a rule such that 38.8.0 users don't get updated to 45.x. Beyond that, I don't think we care how 38.x is treated.

Comment 5

[Rail Aliiev [:rail]]

I bumped the ration to 100% (rules 332) on the release channel and deleted ESR38 rules on release, release-cdntest and release-localtest.

Comment 6

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

(In reply to Wayne Mery (:wsmwk, NI for questions) from comment #4)
> good question. I only recall there should be a rule such that 38.8.0 users
> don't get updated to 45.x. Beyond that, I don't think we care how 38.x is
> treated.

We are not doing this right now, everything 38.0.5 or higher goes 45.3.0.

Comment 7

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

Correction: versions less than 38.5.0 have to go to 38.5.0 for the SHA1 watershed. Everything >= 38.0.5 goes to 45.3.0 (aka the latest release). Please reopen if that's not what you need.

Comment 8

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

Correction: versions less than 38.5.0 have to go to 38.5.0 for the SHA1 watershed. Everything >= 38.5.0 goes to 45.3.0 (aka the latest release). Please reopen if that's not what you need.

Comment 9

[Wayne Mery (:wsmwk)]

Thanks.
I just got the update prompt