Wednesday Triage Report (2020-10-28)

[Sergio Durigan Junior]

On Thursday, October 29 2020, Paride Legovini wrote:

> Sergio Durigan Junior wrote on 28/10/2020:
>> On Wednesday, October 28 2020, Robie Basak wrote:
>>
>>> Hi Sergio,
>>>
>>> On Wed, Oct 28, 2020 at 04:16:37PM -0400, Sergio Durigan Junior wrote:
>>>> ### https://pad.lv/1901708 - (Confirmed) [mysql-8.0] - package
>>>>      mysql-server-8.0 8.0.22-0ubuntu0.20.04.2 failed to install/upgrade:
>>>>      installed mysql-server-8.0 package post-installation script subprocess
>>>>      returned error exit status 1
>>>>
>>>> This seems to be a valid one.  An upgrade from 8.0.21-0ubuntu0.20.04.4
>>>> to 8.0.22-0ubuntu0.20.04.2 will likely trigger a timeout that will
>>>> prevent apt upgrade from finishing, because it requires an internal
>>>> server upgrade that can take several minutes (apparently depending on
>>>> the database size; I couldn't reproduce it locally with an empty db).  I
>>>> subscribe Ubuntu Server, but still marked the bug as Incomplete since
>>>> I'd like to confirm my suspicion that the reporter does have a sizeable
>>>> MySQL database.
>>>
>>> This concern sounds like
>>> https://bugs.launchpad.net/ubuntu/+source/mysql-8.0/+bug/1882527
>>>
>>> Although based on the reporter's reply, perhaps in this case it wasn't
>>> caused by the database being large, and was something else.
>>
>> Hey Robie,
>>
>> Thanks for the reply.  I had a gut feeling that some of you might know a
>> similar issue.  Indeed, it seems like they're related; during my
>> investigations I also noticed the TimeoutSec parameter in the service
>> file, and thought that the next step of the investigation could start
>> there.
>>
>> I left a question to the reporter asking whether they have a big
>> database.  If they do, then I (or someone) can try to populate the test
>> database with dummy entries and try to replicate the scenario.  Let's
>> see.
>
> The bug submitter commented that he didn't have a big database and
> that the system load wasn't particularly high during the
> upgrade. Moreover we have a new almost identical bug:
>
> LP: #1901956 - (New)            [mysql-8.0]      - package
> mysql-server-8.0 8.0.22-0ubuntu0.20.04.2 failed to install/upgrade:
> installed mysql-server-8.0 package post-installation script subprocess
> returned error exit status 1
>
> I asked if there is a big database in place or an unusually high
> system load, but I think we have an actual issue here. What has to be
> understood is if the bug is with the "server upgrade" operation taking
> to long when it shouldn't, or "just" with the timeout, too low for an
> operation that can take several minutes for good reasons.

Thanks, Paride.

Well, the "big database" was more of a hunch, anyway.  I was intrigued
because, on my LXD test container, the "server upgrade" operation only
took a few seconds.  Aside from "big database" or "system under load", I
don't know offhand what might be cause this issue.

I agree with you, we need to determine what makes the "server upgrade"
take so long (almost 5 minutes in one case, and more than 7 minutes in
the other case I analysed), and then understand if this is simply a
scenario where an increase in timeout will be enough, or if there's some
underlying bug.

Thanks,

-- 
Sergio
GPG key ID: E92F D0B3 6B14 F1F4 D8E0  EB2F 106D A1C8 C3CB BF14