Replication Craziness
mischa z
mischa.z at gmail.com
Tue Jun 26 16:07:30 UTC 2007
mogadm was in the path (mogadm: /usr/local/bin/mogadm) - but that didn't
help.
It worked after adding the location to sub _mogadm_exe in
t/lib/mogtestlib.pl.
Also trying to run the test requires shutting down the current live tracker
(that's ok, we have a slave running, but wouldn't it be better to use
different ports from the default ones for testing?)
So here is what we've got:
make test
PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e"
"test_harness(0, 'blib/lib', 'blib/arch')" t/*.t
t/00-startup................ok 3/55Error: Crashed on purpose at
lib/MogileFS/Worker/Query.pm line 176.
t/00-startup................ok 8/55Domain not found.
t/00-startup................ok 18/55Error deleting domain: domain_has_files
Domain still has files, uanble to delete
t/00-startup................ok 24/55# file1 still only on 1 devices
t/00-startup................ok 25/55# Creating 100 files...
# created 10/100
# created 20/100
# created 30/100
# created 40/100
# created 50/100
# created 60/100
# created 70/100
# created 80/100
# created 90/100
# created 100/100
# Files to replicate: 61
t/00-startup................ok 43/55# Files to replicate: 17
t/00-startup................ok 44/55Failure deleting host: host_not_empty
Unable to delete host; it contains devices still
t/00-startup................ok 50/55# Replication update: dev1: 51, dev2:
49, dev3: 49, dev4: 51, dev5: 0, dev6: 0
# Replication update: dev1: 51, dev2: 49, dev3: 0, dev4: 0, dev5: 27,
dev6: 42
t/00-startup................ok 52/55# files on host 1 = 100
# files on host 3 = 69
# files on host 1 = 100
# files on host 3 = 69
# files on host 1 = 100
# files on host 3 = 69
# files on host 1 = 0
# files on host 3 = 0
t/00-startup................ok
t/10-weighting..............ok 10/16Use of uninitialized value in hash
element at lib/MogileFS/ProcManager.pm line 429.
Use of uninitialized value in hash element at lib/MogileFS/ProcManager.pm
line 429.
MogileFS::Backend: tracker socket never became readable (127.0.0.1:7001)
when sending command: [create_open
domain=testdom&fid=0&class=2copies&multi_dest=1&key=file1
] at /usr/local/share/perl/5.8.8/MogileFS/Client.pm line 255
# Looks like you planned 16 tests but only ran 10.
# Looks like your test died just after 10.
t/10-weighting..............dubious
Test returned status 255 (wstat 65280, 0xff00)
DIED. FAILED tests 11-16
Failed 6/16 tests, 62.50% okay
t/20-filepaths..............skipped
all skipped: Filepaths plugin has been seperated from the server, a
bit of work is needed to make the tests run again.
t/domains-classes...........ok
t/fid-stat..................ok
t/hosts-devices.............ok
t/multiple-hosts-replpol....ok
t/store.....................ok
t/util......................ok
Failed Test Stat Wstat Total Fail Failed List of Failed
-------------------------------------------------------------------------------
t/10-weighting.t 255 65280 16 12 75.00% 11-16
1 test skipped.
Failed 1/9 test scripts, 88.89% okay. 6/152 subtests failed, 96.05% okay.
make: *** [test_dynamic] Error 255
On 6/25/07, Brad Fitzpatrick <brad at danga.com> wrote:
>
> 3 tests skipped means pretty much all tests skipped... the 00-startup.t
> test, despite its name, is actually the most important one that tests
> almost everything. You need to get mogadm in the path so it can proceed
> with testing.
>
>
> On Thu, 21 Jun 2007, mischa z wrote:
>
> > Hi, I will try to answer this since Russ is not in the office at the
> moment.
> >
> > How big/small are these files?
> >
> >
> > 4MB in average
> >
> > Anything interesting/noisy in the
> > > mogilefsd error logs?
> >
> >
> > Looks all good. Occasional there are timeouts:
> > " :: [monitor(11947)] Timeout contacting machine 10.0.10.24 for dev 47:
> > took 2.00 seconds out of 2 allowed"
> >
> > Does the mogilefs-server "make test" pass for you?
> >
> >
> > "All tests successful, 3 tests skipped."
> > On the nodes the test skips the Db tests and it complains about not
> finding
> > the mogadm executable. Nothing crucial I guess.
> >
> > Which MySQL server version?
> >
> >
> > mysql> SELECT version();
> > +------------+
> > | version() |
> > +------------+
> > | 5.0.37-log |
> > +------------+
> >
> > Cheers,
> > Mischa
> >
> > On Wed, 20 Jun 2007, Russell Garrett wrote:
> > >
> > > > OK, so I totally wiped the file_to_replicate table. We have 3 nodes
> > > > online with sufficient space. I start the replicate thread and this
> > > happens:
> > > >
> > > > mysql> select * from file_on where fid = 2534860;
> > > > +---------+-------+
> > > > | fid | devid |
> > > > +---------+-------+
> > > > | 2534860 | 54 |
> > > > | 2534860 | 64 |
> > > > | 2534860 | 73 |
> > > > | 2534860 | 74 |
> > > > +---------+-------+
> > > > 4 rows in set (0.00 sec)
> > > >
> > > > ...30 seconds later:
> > > >
> > > > mysql> select * from file_on where fid = 2534860;
> > > > +---------+-------+
> > > > | fid | devid |
> > > > +---------+-------+
> > > > | 2534860 | 54 |
> > > > | 2534860 | 55 |
> > > > | 2534860 | 56 |
> > > > | 2534860 | 57 |
> > > > | 2534860 | 58 |
> > > > | 2534860 | 59 |
> > > > | 2534860 | 60 |
> > > > | 2534860 | 61 |
> > > > | 2534860 | 62 |
> > > > | 2534860 | 63 |
> > > > | 2534860 | 64 |
> > > > | 2534860 | 65 |
> > > > | 2534860 | 66 |
> > > > | 2534860 | 67 |
> > > > | 2534860 | 68 |
> > > > | 2534860 | 69 |
> > > > | 2534860 | 70 |
> > > > | 2534860 | 71 |
> > > > | 2534860 | 72 |
> > > > | 2534860 | 73 |
> > > > | 2534860 | 74 |
> > > > | 2534860 | 75 |
> > > > | 2534860 | 76 |
> > > > | 2534860 | 77 |
> > > > | 2534860 | 78 |
> > > > | 2534860 | 79 |
> > > > +---------+-------+
> > > > 26 rows in set (0.00 sec)
> > > >
> > > >
> > > > Riddle me that one...
> > > >
> > > > Russ Garrett
> > > > Last.fm Ltd.
> > > > russ at last.fm
> > > >
> > > >
> > >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.danga.com/pipermail/mogilefs/attachments/20070626/f4ca377f/attachment.htm
More information about the mogilefs
mailing list