infra-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sebb (JIRA)" <>
Subject [jira] [Commented] (INFRA-16058) lists.a.o: id generation changed on around Sep 2 2017
Date Mon, 19 Feb 2018 01:30:00 GMT


Sebb commented on INFRA-16058:

I mean the id generated from the source mail, which is then in the Permalink and the Source
It is also used for the ElasticSearch database id.

The following mail appears to have been unaffected:
(i.e. when I load it into my test db it has the same id)

However other mails since Sep 2 have different ids.

Whilst this does not immediately affect existing published Permalinks, it does imply that
it is not possible to recover from an ES database problem by reloading from the sources as
there is no guarantee that the same id will be generated.

> lists.a.o: id generation changed on around Sep 2 2017
> -----------------------------------------------------
>                 Key: INFRA-16058
>                 URL:
>             Project: Infrastructure
>          Issue Type: Bug
>          Components: Mail Archives
>            Reporter: Sebb
>            Priority: Major
> The id generation algorithm started producing different ids some time around Sep 2nd
> Before this time, mboxes downloaded from minotaur generate the same ids as lists.a.o
when loaded into a test database using the current trunk software.
> After this time, all the ids are different, although they have the same syntax.
> I have checked jmeter-dev/user and commons-dev/user, but I first noticed the problem
when trying to fix:
> When the freemarker-dev mbox for Feb 2018 was loaded into a test database I was unable
to find the test mail initially as it had a different id.
> It's a bit concerning that the generated ids should change.
> Was it a deliberate algorithm change (if so why was the syntax not changed) or was it
an unexpected side effect of some other software change?

This message was sent by Atlassian JIRA

View raw message