Thank you Ellie ! :-)
All tests pass with the master branch on Ubuntu 15.04 now as well.
Best regards,
Conrad Kleinespel
conr...@conradk.com
+33 6 23 82 42 79
On Mon, May 4, 2015, at 08:03 AM, ellie timoney wrote:
> > Will raise a maniphest task for the bug, but in the meantime you can get
>
> Raised and resolved :)
>
> On Mon, May 4, 2015, at 03:42 PM, ellie timoney wrote:
> > Another missing dependency, this time a cyrus one:
> >
> > sudo apt-get install uuid-dev
> >
> > and then rebuild cyrus from scratch.
> >
> > The problem looks like a bug in the lib/util.c:makeuuid(), specifically
> > its handling of the case where we don't HAVE_LIBUUID. The fallback is
> > to generate a uuid from random bytes, which at a glance looks like it
> > should be okay except that for whatever reason it isn't, and results in
> > strings like "t9|ltv0lnout\204z\200\066|~5p2\200ly" (look at all those
> > octals!). This winds up in mailboxes.db as a quoted literal (i.e.
> > {24}\r\n.......), and then dlist_parsesax is not happy when it comes
> > time to look for those mailboxes later.
> >
> > Will raise a maniphest task for the bug, but in the meantime you can get
> > running just by installing uuid-dev. At this point all Cassandane tests
> > pass for master and cyrus-imapd-2.5 branches (... with the occasional
> > exception of Master.sighup_recycling)
> >
> > ellie
> >
> > On Mon, May 4, 2015, at 01:15 PM, ellie timoney wrote:
> > > > This one:
> > > >
> > > > > 4 Perl exception: child process (binary
> > > > > /home/ellie/build/cassandane/utils/annotator.pl pid [pid]) exited
> > > > > with code 255
> > > >
> > > > ... is a head scratcher. I'm digging deeper into it at the moment.
> > >
> > > After all that, this one was just missing runtime dependencies:
> > >
> > > sudo apt-get install libnet-server-perl libunix-syslog-perl
> > >
> > >
> > >
> > > On Mon, May 4, 2015, at 01:02 PM, ellie timoney wrote:
> > > > Okay, here's a summary of the errors I'm currently getting from
> > > > Cassandane against the a cyrus built from the master branch on Debian
> > > > 8.0
> > > >
> > > > > ellie@cyrus-test:~/build/cassandane$ ./testrunner.pl -f pretty -j 8
> > > > > 2>/dev/null | grep -A2 '^[0-9][0-9]*) ' | grep '^ Perl exception:' |
> > > > > sed -e 's/:[0-9][0-9]*$/:[port]/' -e 's/pid [0-9][0-9]*/pid [pid]/' |
> > > > > sort | uniq -c | sort -nrk 1
> > > > > 91 Perl exception: Cannot setacl for user.cassandane: IMAP
> > > > > Command : 'setacl' failed. Response was : no - Invalid identifier
> > > > > 58 Perl exception: Some process is already listening on
> > > > > 127.0.0.1:[port]
> > > > > 6 Perl exception: Cannot setacl for user/cassandane: IMAP
> > > > > Command : 'setacl' failed. Response was : no - Invalid identifier
> > > > > 4 Perl exception: child process (binary
> > > > > /home/ellie/build/cassandane/utils/annotator.pl pid [pid]) exited
> > > > > with code 255
> > > > > 1 Perl exception: Timed out waiting for lemmings to reach the
> > > > > expected census
> > > > > 1 Perl exception: Cannot setmetadata: IMAP Command :
> > > > > 'setmetadata' failed. Response was : no - Mailbox does not exist
> > > > > 1 Perl exception: Cannot setacl for user.gastropub: IMAP
> > > > > Command : 'setacl' failed. Response was : no - Invalid identifier
> > > > >
> > > >
> > > >
> > > > These ones:
> > > >
> > > > > 58 Perl exception: Some process is already listening on
> > > > > 127.0.0.1:[port]
> > > >
> > > > ... I have seen before, and at the time the problem was a few stray
> > > > masters hanging around from a previous cass run that I had ^C'd and not
> > > > cleaned up after manually -- cleaning those up at the time fixed the
> > > > errors. This time that isn't the problem though -- when the Cassandane
> > > > run completes, there's no cyrus processes left running at all. So, so
> > > > far I dunno.
> > > >
> > > > These two:
> > > >
> > > > > 1 Perl exception: Cannot setmetadata: IMAP Command :
> > > > > 'setmetadata' failed. Response was : no - Mailbox does not exist
> > > > > 1 Perl exception: Cannot setacl for user.gastropub: IMAP
> > > > > Command : 'setacl' failed. Response was : no - Invalid identifier
> > > >
> > > > ... seem to be sporadic, they don't seem to occur on every Cassandane
> > > > run, but did on this one.
> > > >
> > > > This one:
> > > >
> > > > > 4 Perl exception: child process (binary
> > > > > /home/ellie/build/cassandane/utils/annotator.pl pid [pid]) exited
> > > > > with code 255
> > > >
> > > > ... is a head scratcher. I'm digging deeper into it at the moment.
> > > >
> > > > These ones:
> > > >
> > > > > 91 Perl exception: Cannot setacl for user.cassandane: IMAP
> > > > > Command : 'setacl' failed. Response was : no - Invalid identifier
> > > > > 6 Perl exception: Cannot setacl for user/cassandane: IMAP
> > > > > Command : 'setacl' failed. Response was : no - Invalid identifier
> > > > > 1 Perl exception: Cannot setacl for user.gastropub: IMAP
> > > > > Command : 'setacl' failed. Response was : no - Invalid identifier
> > > >
> > > > I thought I had seen something like this (specifically, "Cannot
> > > > setacl...") with the master branch on Debian 7.8 a few weeks ago, but
> > > > then I didn't look at master much for a while and next time I did it
> > > > wasn't happening anymore, so I just assumed it was an actual bug on
> > > > master that had gotten fixed. (IIRC it might have gone away when
> > > > Anthony's configure changes were reverted?)
> > > >
> > > > So that's where I'm at. I'll keep digging (specifically into the
> > > > annotator.pl problem for now), but if any of this looks familiar to
> > > > anyone else on the list I'd appreciate a pointer. :)
> > > >
> > > > Cheers,
> > > >
> > > > ellie
> > > >
> > > > On Mon, May 4, 2015, at 10:30 AM, ellie timoney wrote:
> > > > > Hi Conrad,
> > > > >
> > > > > I don't know about Debian 8 currently, but if we're not considering
> > > > > it a
> > > > > supported platform now, I expect we will once Fastmail upgrades to
> > > > > it.
> > > > > But we don't yet have even a vague timeline on when that will be.
> > > > >
> > > > > I don't know much/anything about Docker. My understanding was that
> > > > > the
> > > > > current Docker images were more for testing purposes, and that at some
> > > > > point we would start officially releasing Docker images for people who
> > > > > want to run Cyrus without having to manage a whole system under
> > > > > it(?).
> > > > > But I don't imagine they'd be useful as development environments.
> > > > >
> > > > > I'll keep poking around on my Debian 8 VM as I have time, and see if I
> > > > > can find a fix for the errors from Cassandane.
> > > > >
> > > > > Cheers,
> > > > >
> > > > > ellie
> > > > >
> > > > > On Sat, May 2, 2015, at 08:10 PM, Conrad Kleinespel wrote:
> > > > > > Hello Ellie,
> > > > > >
> > > > > > Thanks for the tip. I just ran the tests on the 2.5 branch. I get a
> > > > > > lot
> > > > > > less errors and no failure:
> > > > > > Run: 252, Failures: 0, Errors: 42
> > > > > >
> > > > > > Seeing that you have the same issue is a relief (kinda). Are Debian
> > > > > > 8
> > > > > > and Ubuntu officially supported platforms ? If not, what I'd do is
> > > > > > just
> > > > > > setup a Docker container with Debian 7.8 and see if I can run
> > > > > > everything
> > > > > > in there. I see there is a Docker repository on Phabricator. Would
> > > > > > that
> > > > > > be a solution or are these Docker containers not to be used for some
> > > > > > reason ? Seeing how you took the time to write all of these
> > > > > > instructions, I'm unsure.
> > > > > >
> > > > > > Thanks for your help,
> > > > > >
> > > > > > Best regards.
> > > > > >
> > > > > > --
> > > > > > Conrad Kleinespel
> > > > > > conr...@conradk.com
> > > > > > +33 6 23 82 42 79
> > > > > >
> > > > > > On Sat, May 2, 2015, at 10:17 AM, ellie timoney wrote:
> > > > > > > > However, when running the tests with "./testrunner.pl -f pretty
> > > > > > > > -j 8", I
> > > > > > > > get 1 failure and 182 errors in 252 tests. That seems odd. I
> > > > > > > > see no
> > > > > > > > specific error messages in syslog or elsewhere, so I'm thinking
> > > > > > > > I must
> > > > > > > > be missing something. Have any of you experienced this ?
> > > > > > >
> > > > > > > Is that against the cyrus master branch? Do you get less errors
> > > > > > > with
> > > > > > > the cyrus-imapd-2.5 branch?
> > > > > > >
> > > > > > > I experienced a lot (don't recall numbers offhand, sorry) of
> > > > > > > errors
> > > > > > > against master, and a lot (but fewer) against cyrus-imapd-2.5,
> > > > > > > when
> > > > > > > running Cassandane on my Debian 8.0 VM, which is probably similar
> > > > > > > to
> > > > > > > recent Ubuntu. But all tests are fine for me on both branches on
> > > > > > > Debian
> > > > > > > 7.8. So I don't yet know if something has changed in recent
> > > > > > > Debian/Ubuntu to cause this, or if I've just missed a step in the
> > > > > > > VM
> > > > > > > setup (and consequently the instructions).
> > > > > > >
> > > > > > > On Sat, May 2, 2015, at 08:20 AM, Conrad Kleinespel wrote:
> > > > > > > > Hello guys and girls,
> > > > > > > >
> > > > > > > > Ellie, thanks a lot for the instructions. Building Cyrus and
> > > > > > > > Cassandane
> > > > > > > > worked just fine on Ubuntu 15.10 :-)
> > > > > > > >
> > > > > > > > However, when running the tests with "./testrunner.pl -f pretty
> > > > > > > > -j 8", I
> > > > > > > > get 1 failure and 182 errors in 252 tests. That seems odd. I
> > > > > > > > see no
> > > > > > > > specific error messages in syslog or elsewhere, so I'm thinking
> > > > > > > > I must
> > > > > > > > be missing something. Have any of you experienced this ?
> > > > > > > >
> > > > > > > > Installing arcanist on Ubuntu was apparently as easy as "sudo
> > > > > > > > apt-get
> > > > > > > > install arcanist". I'll see about actually using it together
> > > > > > > > with
> > > > > > > > Phabricator when I make my first commit.
> > > > > > > >
> > > > > > > > Best regards,
> > > > > > > >
> > > > > > > > Conrad Kleinespel
> > > > > > > > conr...@conradk.com
