Merge lp:~hjd/apport/runit-bugpattern into lp:~ubuntu-bugcontrol/apport/ubuntu-bugpatterns

[Hans Joachim Desserud]

Bug pattern for bug 1448164. There's a somewhat steady stream of new duplicates for this issue and unfortunately, it is a Won't Fix for 16.04.

Couple of questions:
 1. I don't know how strict the Package filtering is. I've seen some of the duplicated being reported to git, due to git-daemon-run's dependency on runit. Should the package filter be expanded to also check for this package name?
 2. Did some testing with test-local, and discovered that the master bug doesn't seem to have any apport information. Is this an issue?
 3. Some of the early package versions in Ubuntu 16.10, but _not_ the current one also had this issue. Should the pattern for version number be expanded to cover these as well?

Otherwise, see commit messages for details :)

PS. Please disregard/close the original merge request (https://code.launchpad.net/~hjd/apport/runit-bugpattern/+merge/315404)

[Martin Pitt (pitti)]

> PS. Please disregard/close the original merge request (https://code.launchpad.net/~hjd/apport/runit-bugpattern/+merge/315404)

Deleted :-)

[Alberto Salvia Novella (es20490446e)]

review: Approve

[Brian Murray (brian-murray)]

1. I'm not sure what you mean by "how strict it is", but expanding it to check the other package makes sense to me given that the Package key is a regular expression and it only currently catch bugs about runit.

2. It's sort of an issue in that test-local will stop processing if the bug doesn't have apport data, and test-local has the ability to check duplicates of a master bug. So you lose out on some of the powers of test-local, but its not a blocking issue.

3. If I understand correctly it's unlikely that we'll be receiving new duplicates with the previous version of the package in 16.10 so the pattern wouldn't help there. However, if you used consolidate-bugs with a pattern with those previous package versions you could help clean up old duplicates so you might add the previous versions and use consolidate-bugs.

Thanks for working on this!

review: Needs Fixing

[lp:~hjd/apport/runit-bugpattern]

571. By Hans Joachim Desserud on 2017-01-25

Escape forward slashes

[Hans Joachim Desserud (hjd)]

Good catch with the forward slashes, I totally forgot.
I note that test-local report a match both with and without, but it seems better to have them there.

1. Sorry for being unclear. I assume the Package key is used to match the package where the error occured to check if it is the same or just a similar problem in a different package, but I am not sure how it decides which package name should be check.
The issue occurs when installing runit, so if I install that package alone, the error triggers and it will match the Package key. In this case the error originates from the package, and it's a pretty straight-forward case. If I instead attempt to install git-daemon-run, this will pull in runit as a dependency and fail as well. Will it then try to match runit or git-daemon-run against the Package key? Because the error always occurs in runit, but it might be triggered when installing other packages that depend on it, should those then be listed as alternatives?

2. Ok, as long as the real-world use won't have similar problems, I can live with that. I tested manually with some of the other duplicates.

3. That's a good point. Though at least for runit directly, I've marked the duplicates manually. The remaining bugs there seemed to be either unrelated or with different error messages, so I'm not sure what's up with those.

[Brian Murray (brian-murray)]

On Wed, Jan 25, 2017 at 06:03:02PM -0000, Hans Joachim Desserud wrote:
> Good catch with the forward slashes, I totally forgot.
> I note that test-local report a match both with and without, but it seems better to have them there.
>
> 1. Sorry for being unclear. I assume the Package key is used to match the package where the error occured to check if it is the same or just a similar problem in a different package, but I am not sure how it decides which package name should be check.

The "Package" key is searched for in the bug description e.g. "Package:
git-daemon-run 1:2.7.4-0ubuntu1" from LP: #1625798.

> The issue occurs when installing runit, so if I install that package
> alone, the error triggers and it will match the Package key. In this
> case the error originates from the package, and it's a pretty
> straight-forward case. If I instead attempt to install git-daemon-run,
> this will pull in runit as a dependency and fail as well. Will it then
> try to match runit or git-daemon-run against the Package key?

git-daemon-run because that's what'll be in the Package field for the
report.

> Because the error always occurs in runit, but it might be triggered
> when installing other packages that depend on it, should those then be
> listed as alternatives?

Yes, I think so. You might use "apt-cache rdepends runit" and check
those packages for bugs like this too.

> 2. Ok, as long as the real-world use won't have similar problems, I
> can live with that. I tested manually with some of the other
> duplicates.
>
> 3. That's a good point. Though at least for runit directly, I've
> marked the duplicates manually. The remaining bugs there seemed to be
> either unrelated or with different error messages, so I'm not sure
> what's up with those.

--
Brian Murray
Ubuntu Bug Master

[lp:~hjd/apport/runit-bugpattern]

572. By Hans Joachim Desserud on 2017-01-31

Added rdepends of runit to bugpattern

[Hans Joachim Desserud (hjd)]

>The "Package" key is searched for in the bug description
Oooh, that "Package" :)

Ok, added the rdepends which depend directly on runit now.

=== modified file 'bugpatterns.xml'
--- bugpatterns.xml	2017-01-13 19:03:11 +0000
+++ bugpatterns.xml	2017-01-31 18:19:24 +0000
@@ -1121,6 +1121,14 @@
 	<re key="Traceback">authorizer = OAuthAuthorizer\(token_key=credential\['token'\],</re>
     </pattern>
 
+<!-- openjdk-9 -->
+
+    <pattern url="https://launchpad.net/bugs/1550950">
+        <re key="Package">^openjdk-9-jdk</re>
+        <re key="DpkgTerminalLog">openjdk-9-(jdk|jdk-headless):.* \(9~(b107-0ubuntu1|b112-2|b114-0ubuntu1)\)</re>
+        <re key="DpkgTerminalLog">include\/linux\/jawt_md.h.*openjdk-9-jdk-headless</re>
+    </pattern>
+
 <!-- Converted from openswan.xml -->
     <pattern url="https://launchpad.net/bugs/739001">
         <re key="Package">^openswan-modules-dkms </re>
@@ -1182,6 +1190,14 @@
 	<re key="StacktraceTop">g_object_get_valist</re>
     </pattern>
 
+<!-- Applies to runit -->
+
+    <pattern url="https://launchpad.net/bugs/1448164">
+        <re key="Package">^(runit |bcron-run |vblade-persist |twoftpd-run |socklog-run |runit-upstart |getty-run |runit-systemd |runit-init |qmail-run |mongrel2-run |git-daemon-run |cereal |dnscache-run )</re>
+        <re key="DpkgTerminalLog">runit \(2.1.2-3ubuntu1\)</re>
+        <re key="DpkgTerminalLog">start:.*Upstart: Failed to connect to socket \/com\/ubuntu\/upstart:.*</re>
+    </pattern>
+
 <!-- Applies to samba -->
 
     <pattern url="https://launchpad.net/bugs/877852">
@@ -2115,10 +2131,5 @@
         <re key="modified.conffile..etc.mysql.my.cnf.fallback">\[deleted\]</re>
     </pattern>
 
-    <pattern url="https://launchpad.net/bugs/1550950">
-        <re key="Package">^openjdk-9-jdk</re>
-        <re key="DpkgTerminalLog">openjdk-9-(jdk|jdk-headless):.* \(9~(b107-0ubuntu1|b112-2|b114-0ubuntu1)\)</re>
-        <re key="DpkgTerminalLog">include\/linux\/jawt_md.h.*openjdk-9-jdk-headless</re>
-    </pattern>
 
 </patterns>