The error "ERROR:  could not read block 326 in file "base/257536/259291": read only 0 of 8192 bytes" line makes me wonder if your database may have been corrupt prior to the upgrade/dump.  Some brief googling suggests that this could be hardware failure induced among other things.  Can you check the postgres logs for any likely explanations?

How exactly did you perform your database backup and attempted recreation?

-Dannon


On Mon, Jun 17, 2013 at 5:12 PM, Leon Mei <hailiang.mei@nbic.nl> wrote:
Dear list,

Yesterday, I noticed that the user quota setting on our Galaxy had a problem - the list of user email address is missing from the manage quota page. So I thought to give a Galaxy codebase update a try after backing up the Galaxy dir and the database.

Running hg head gives me:
==
postgres@ubuntu:/mnt/galaxyTools/galaxy-2013-05-13$ hg head
changeset:   10003:b4a373d86c51
tag:         tip
parent:      10001:471484ff8be6
user:        greg
date:        Wed Jun 12 11:48:09 2013 -0400
summary:     Add targets to Repository Actions menu items.
==

I updated the DB schema to support the code update. But the restart of Galaxy failed with an error in paster.log:
==
Traceback (most recent call last):
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/webapps/galaxy/buildapp.py", line 35, in app_factory
    app = UniverseApplication( global_conf = global_conf, **kwargs )
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/app.py", line 164, in __init__
    self.job_manager = manager.JobManager( self )
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/jobs/manager.py", line 36, in __init__
    self.job_handler.start()
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/jobs/handler.py", line 34, in start
    self.job_queue.start()
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/jobs/handler.py", line 77, in start
    self.__check_jobs_at_startup()
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/jobs/handler.py", line 125, in __check_jobs_at_startup
    self.dispatcher.recover( job, job_wrapper )
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/jobs/handler.py", line 620, in recover
    self.job_runners[runner_name].recover( job, job_wrapper )
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/jobs/runners/local.py", line 128, in recover
    job_wrapper.change_state( model.Job.states.ERROR, info = "This job was killed when Galaxy was restarted.  Please retry the job." )
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/jobs/__init__.py", line 824, in change_state
    dataset.state = state
  File "/mnt/galaxyTools/galaxy-2013-05-13/lib/galaxy/model/__init__.py", line 1163, in set_dataset_state
    object_session( self ).flush() #flush here, because hda.flush() won't flush the Dataset object
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/orm/session.py", line 1718, in flush
    self._flush(objects)
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/orm/session.py", line 1789, in _flush
    flush_context.execute()
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/orm/unitofwork.py", line 331, in execute
    rec.execute(self)
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/orm/unitofwork.py", line 475, in execute
    uow
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/orm/persistence.py", line 59, in save_obj
    mapper, table, update)
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/orm/persistence.py", line 485, in _emit_update_statements
    execute(statement, params)
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/engine/base.py", line 1449, in execute
    params)
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/engine/base.py", line 1584, in _execute_clauseelement
    compiled_sql, distilled_params
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/engine/base.py", line 1698, in _execute_context
    context)
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/engine/base.py", line 1691, in _execute_context
    context)
  File "/mnt/galaxyTools/galaxy-2013-05-13/eggs/SQLAlchemy-0.7.9-py2.7-linux-x86_64-ucs4.egg/sqlalchemy/engine/default.py", line 331, in do_execute
    cursor.execute(statement, parameters)
InternalError: (InternalError) could not read block 326 in file "base/257536/259291": read only 0 of 8192 bytes
 'UPDATE dataset SET update_time=%(update_time)s, state=%(state)s WHERE dataset.id = %(dataset_id)s' {'update_time': datetime.datetime(2013, 6, 17, 12, 42, 38, 145908), 'state': 'error', 'dataset_id': 68792}
==

I went on to check the Galaxy database (in postgres) which confirmed the error in the paster.log.
==
galaxy=# select * from dataset where id = 68792;                                                                                                                                                               id   |        create_time         |        update_time         | state | deleted | purged | purgable | external_filename | _extra_files_path | file_size | total_size | object_store_id | uuid
-------+----------------------------+----------------------------+-------+---------+--------+----------+-------------------+-------------------+-----------+------------+-----------------+------
 68792 | 2013-06-12 06:55:17.608725 | 2013-06-12 06:55:19.396557 | error | f       | f      | t        |                   |                   |           |            |                 |
(1 row)

galaxy=# UPDATE dataset SET update_time='2013-06-17 12:42:38.145908', state='error' WHERE dataset.id = 68792;
ERROR:  could not read block 326 in file "base/257536/259291": read only 0 of 8192 bytes
galaxy=# UPDATE dataset SET state='error' WHERE dataset.id = 68792;
UPDATE 1
==

When I tried to restore the database from the dumped copy I made yesterday, I got the following error:
==
ERROR:  insert or update on table "galaxy_session_to_history" violates foreign key constraint "galaxy_session_to_history_
session_id_fkey"
DETAIL:  Key (session_id)=(956338) is not present in table "galaxy_session".
ERROR:  insert or update on table "job" violates foreign key constraint "job_session_id_fkey"
DETAIL:  Key (session_id)=(852060) is not present in table "galaxy_session".
ERROR:  insert or update on table "workflow_step_connection" violates foreign key constraint "workflow_step_connection_input_step_id_fkey"
DETAIL:  Key (input_step_id)=(18422) is not present in table "workflow_step".
ERROR:  insert or update on table "workflow_step_connection" violates foreign key constraint "workflow_step_connection_output_step_id_fkey"
DETAIL:  Key (output_step_id)=(18422) is not present in table "workflow_step".
==

It seems our postgres DB got corrupted? Any suggestion on a fix?

Thanks a lot!

Leon

--
Hailiang (Leon) Mei
Netherlands Bioinformatics Center  
BioAssist NGS Taskforce
 - http://ngs.nbic.nl
Skype: leon_mei    Mobile: +31 6 41709231

___________________________________________________________
Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:
  http://lists.bx.psu.edu/

To search Galaxy mailing lists use the unified search at:
  http://galaxyproject.org/search/mailinglists/