Discussion:
Rebuilding the repository
m***@mherrn.de
2016-09-22 15:13:55 UTC
Permalink
Hello,

I see lots of problems in the archiva.log that the process cannot find
some bundles:

2016-09-22 17:08:02,435 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
[] - failed to read bundle: b9706fc7-846b-4dab-bac3-7a734136dbf8:
java.io.EOFException
2016-09-22 17:08:02,477 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
[] - failed to read bundle: b9706fc7-846b-4dab-bac3-7a734136dbf8:
java.io.EOFException
2016-09-22 17:08:02,528 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
[] - failed to read bundle: b9706fc7-846b-4dab-bac3-7a734136dbf8:
java.io.EOFException
2016-09-22 17:08:02,549 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
[] - failed to read bundle: 27c8fb7c-b9ef-4b6c-9b59-565d71f51ce5:
java.io.EOFException


I expect this happened because some files were removed directly in the
filesystem.


This leads to two questions:

1. How can I recreate the whole repository from scratch? I tried
configuring a different location in archiva.xml, but that lead to the same
problems.

2. How can archiva be made a bit more robust? Files or directories being
deleted directly in the filesystem shouldn't confuse archiva that much
that it doesn't really revive.

Best regards
Marco
Martin Stockhammer
2016-09-23 06:48:44 UTC
Permalink
Hi Marco,

a rescan (directories scanning action) does not help?


For 2: would be great if you can create a issue in the Jira.

Greetings

Martin
Post by m***@mherrn.de
Hello,
I see lots of problems in the archiva.log that the process cannot find
2016-09-22 17:08:02,435 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
java.io.EOFException
2016-09-22 17:08:02,477 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
java.io.EOFException
2016-09-22 17:08:02,528 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
java.io.EOFException
2016-09-22 17:08:02,549 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
java.io.EOFException
I expect this happened because some files were removed directly in the
filesystem.
1. How can I recreate the whole repository from scratch? I tried
configuring a different location in archiva.xml, but that lead to the same
problems.
2. How can archiva be made a bit more robust? Files or directories being
deleted directly in the filesystem shouldn't confuse archiva that much
that it doesn't really revive.
Best regards
Marco
--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail gesendet.
m***@mherrn.de
2016-09-28 10:46:15 UTC
Permalink
Hi Martin,
Post by Martin Stockhammer
a rescan (directories scanning action) does not help?
I have done that and now the server has massive load since days (the IO
connection is unfortunately very bad, so that may be the main cause).
However the problems about missing bundles still persists.
I have to recheck this with a new empty repository.

If this doesn't help, do you have an idea on how to best recreate the
repostory then?
Post by Martin Stockhammer
For 2: would be great if you can create a issue in the Jira.
OK, I will do that.

Best regards
Marco
Olivier Lamy
2016-09-29 01:13:05 UTC
Permalink
Hey
Did you delete all files from the repo or only few?
I would delete all otherwise the jackrabbit db will be really corrupted!
Post by m***@mherrn.de
Hi Martin,
Post by Martin Stockhammer
a rescan (directories scanning action) does not help?
I have done that and now the server has massive load since days (the IO
connection is unfortunately very bad, so that may be the main cause).
However the problems about missing bundles still persists.
I have to recheck this with a new empty repository.
If this doesn't help, do you have an idea on how to best recreate the
repostory then?
Post by Martin Stockhammer
For 2: would be great if you can create a issue in the Jira.
OK, I will do that.
Best regards
Marco
--
Olivier Lamy
http://twitter.com/olamy | http://linkedin.com/in/olamy
m***@mherrn.de
2016-09-29 10:50:10 UTC
Permalink
Hi Olivier,

only some of them were deleted.
I will try to remove all of them and if this fixes the problem.

Best regards
Marco
Post by Olivier Lamy
Hey
Did you delete all files from the repo or only few?
I would delete all otherwise the jackrabbit db will be really corrupted!
Post by m***@mherrn.de
Hi Martin,
Post by Martin Stockhammer
a rescan (directories scanning action) does not help?
I have done that and now the server has massive load since days (the IO
connection is unfortunately very bad, so that may be the main cause).
However the problems about missing bundles still persists.
I have to recheck this with a new empty repository.
If this doesn't help, do you have an idea on how to best recreate the
repostory then?
Post by Martin Stockhammer
For 2: would be great if you can create a issue in the Jira.
OK, I will do that.
Best regards
Marco
--
Olivier Lamy
http://twitter.com/olamy | http://linkedin.com/in/olamy
m***@mherrn.de
2016-10-10 15:28:14 UTC
Permalink
Hello all,

I have now had the time to try to fix my problem.
I have stopped archiva,
then I have modified the archiva.xml to point the internal and snapshot
repositories to totally different (empty) directories.
Then I have started archiva up again.
And did a directory scan for both repositories.
However I still get lots of

2016-10-10 17:23:48,401 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
[] - failed to read bundle: ed7c5e19-3418-4ab1-a810-a955469969d1:
java.io.EOFException

message in the log file. And they don't stop. Archiva is writing them on
and on..

Also when browing the web UI, I get the message

failed to retrieve item state of item
51a07434-8277-41a2-80ad-61d8415b8642

for the groupId I have snapshot versions of.

How can I get the database back into a consistent state?


Best regards
Marco
Post by m***@mherrn.de
Hi Olivier,
only some of them were deleted.
I will try to remove all of them and if this fixes the problem.
Best regards
Marco
Post by Olivier Lamy
Hey
Did you delete all files from the repo or only few?
I would delete all otherwise the jackrabbit db will be really corrupted!
Post by m***@mherrn.de
Hi Martin,
Post by Martin Stockhammer
a rescan (directories scanning action) does not help?
I have done that and now the server has massive load since days (the IO
connection is unfortunately very bad, so that may be the main cause).
However the problems about missing bundles still persists.
I have to recheck this with a new empty repository.
If this doesn't help, do you have an idea on how to best recreate the
repostory then?
Post by Martin Stockhammer
For 2: would be great if you can create a issue in the Jira.
OK, I will do that.
Best regards
Marco
--
Olivier Lamy
http://twitter.com/olamy | http://linkedin.com/in/olamy
Martin
2016-10-16 17:44:10 UTC
Permalink
Hi Marco,

unfortunately the BundleFsPersistenceManager has no option to run a
consistency check during startup. With the current archiva I see no way to
force a consistency check on the running repository.

If you use the standalone archiva, one way could be:
stop archiva
Backup <archiva_home>/repositories
Remove <archiva_home>/repositories completely
start archiva
Rescan Directories

I'm not sure, if this works and/or if this leads to other errors, but at least
with a out of the box archiva standalone the repositories directory is rebuilt
after the restart.

Greetings

Martin
Post by m***@mherrn.de
Hello all,
I have now had the time to try to fix my problem.
I have stopped archiva,
then I have modified the archiva.xml to point the internal and snapshot
repositories to totally different (empty) directories.
Then I have started archiva up again.
And did a directory scan for both repositories.
However I still get lots of
2016-10-10 17:23:48,401 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
java.io.EOFException
message in the log file. And they don't stop. Archiva is writing them on
and on..
Also when browing the web UI, I get the message
failed to retrieve item state of item
51a07434-8277-41a2-80ad-61d8415b8642
for the groupId I have snapshot versions of.
How can I get the database back into a consistent state?
Best regards
Marco
Post by m***@mherrn.de
Hi Olivier,
only some of them were deleted.
I will try to remove all of them and if this fixes the problem.
Best regards
Marco
Post by Olivier Lamy
Hey
Did you delete all files from the repo or only few?
I would delete all otherwise the jackrabbit db will be really corrupted!
Post by m***@mherrn.de
Hi Martin,
Post by Martin Stockhammer
a rescan (directories scanning action) does not help?
I have done that and now the server has massive load since days (the IO
connection is unfortunately very bad, so that may be the main cause).
However the problems about missing bundles still persists.
I have to recheck this with a new empty repository.
If this doesn't help, do you have an idea on how to best recreate the
repostory then?
Post by Martin Stockhammer
For 2: would be great if you can create a issue in the Jira.
OK, I will do that.
Best regards
Marco
--
Olivier Lamy
http://twitter.com/olamy | http://linkedin.com/in/olamy
m***@mherrn.de
2016-10-17 06:26:23 UTC
Permalink
Hi Martin,

I tried removing all repositories and then starting up archiva again, but
that lead to even more probolems. Nothing could be found in the repository
anymore and still there were lots of errors in the log. Rescanning
directories didn't help either. I had to do a totally new installation of
archiva to get rid of the errors.

Best regards
Marco
Post by Martin Stockhammer
Hi Marco,
unfortunately the BundleFsPersistenceManager has no option to run a
consistency check during startup. With the current archiva I see no way to
force a consistency check on the running repository.
stop archiva
Backup <archiva_home>/repositories
Remove <archiva_home>/repositories completely
start archiva
Rescan Directories
I'm not sure, if this works and/or if this leads to other errors, but at least
with a out of the box archiva standalone the repositories directory is rebuilt
after the restart.
Greetings
Martin
Post by m***@mherrn.de
Hello all,
I have now had the time to try to fix my problem.
I have stopped archiva,
then I have modified the archiva.xml to point the internal and snapshot
repositories to totally different (empty) directories.
Then I have started archiva up again.
And did a directory scan for both repositories.
However I still get lots of
2016-10-10 17:23:48,401 [pool-6-thread-1] ERROR
org.apache.jackrabbit.core.persistence.bundle.BundleFsPersistenceManager
java.io.EOFException
message in the log file. And they don't stop. Archiva is writing them on
and on..
Also when browing the web UI, I get the message
failed to retrieve item state of item
51a07434-8277-41a2-80ad-61d8415b8642
for the groupId I have snapshot versions of.
How can I get the database back into a consistent state?
Best regards
Marco
Post by m***@mherrn.de
Hi Olivier,
only some of them were deleted.
I will try to remove all of them and if this fixes the problem.
Best regards
Marco
Post by Olivier Lamy
Hey
Did you delete all files from the repo or only few?
I would delete all otherwise the jackrabbit db will be really
corrupted!
Post by m***@mherrn.de
Post by Olivier Lamy
Post by m***@mherrn.de
Hi Martin,
Post by Martin Stockhammer
a rescan (directories scanning action) does not help?
I have done that and now the server has massive load since days (the
IO
Post by m***@mherrn.de
Post by Olivier Lamy
Post by m***@mherrn.de
connection is unfortunately very bad, so that may be the main
cause).
Post by m***@mherrn.de
Post by Olivier Lamy
Post by m***@mherrn.de
However the problems about missing bundles still persists.
I have to recheck this with a new empty repository.
If this doesn't help, do you have an idea on how to best recreate
the
Post by m***@mherrn.de
Post by Olivier Lamy
Post by m***@mherrn.de
repostory then?
Post by Martin Stockhammer
For 2: would be great if you can create a issue in the Jira.
OK, I will do that.
Best regards
Marco
--
Olivier Lamy
http://twitter.com/olamy | http://linkedin.com/in/olamy
Loading...