[Pulp-list] Can't get CentOS 7 Updates repo published

Fri Oct 2 11:39:03 UTC 2020

Filelists are stored in the database in a compressed form[0]. During
publishing of metadata, each fillist is decompressed before being written
to a file[1]. It's hard to speculate when data corruption occurred. It's
possible that the data was corrupted in the remote repository the first
time that this RPM was downloaded. That metadata could have been fixed
since then, but Pulp thinks it already has the information and is not
trying to re-download.

[0]
https://github.com/pulp/pulp_rpm/blob/2-master/plugins/pulp_rpm/plugins/db/models.py#L1051
[1]
https://github.com/pulp/pulp_rpm/blob/2-master/plugins/pulp_rpm/plugins/db/models.py#L1064

On Fri, Oct 2, 2020 at 3:27 AM Konstantin M. Khankin <
khankin.konstantin at gmail.com> wrote:

> Sorry for reviving this thread but I think I found something.
>
> Today I tried to install CentOS 7 and got an error:
> "TypeError: Parsing filelists.xml error: expected '>'"
>
> So I did this:
>
> >>> import xml.etree.ElementTree as ET
> >>> tree =
> ET.parse('/mnt/sysimage/var/tmp/yum.cache/centos_updates/gen/filelists.xml')
> ...
>   File "/usr/lib64/python2.7/xml/etree/ElementTree.py", line 1506, in
> _raiseerror
>     raise err
> xml.etree.ElementTree.ParseError: not well-formed (invalid token): line
> 1884123, column 136
>
> I opened filelists.xml and found that line 1884123 is indeed corrupt and
> looks like this:
>
> <file>/usr/share/javadoc/java-1.8.0-openjdk-1.8.0.111-1.b15.el7_2-debug/api/javax/sql/rowset/serial/compact3-package-frame.html</fil%
>   !,^@ 0
> 480!share/javadoc/java-1.8.0-openjdk-1.8.0.111-1.b15.el7_2-debug/api/javax/sql/rowset/serial/compact3-package-summary.html</file>
>
> I verified another consumer has exactly the same issue with this file.
>
> Do I understand correctly Pulp generates filelists.xml? If so, can it be
> because of a bug in Pulp or should I look for a silent data corruption
> issue on my pulp server?
>
> Thanks!
>
>
> вт, 1 сент. 2020 г. в 17:14, Dennis Kliban <dkliban at redhat.com>:
>
>> The repoview feature requires the sqlite db to be generated. This
>> feature generates HTML pages for browsing the repository in a web browser.
>> I believe that users are still able to browse the repository without it,
>> but in that case the HTML listing of directories is generated by the web
>> server each time a user requests it.
>>
>> On Tue, Sep 1, 2020 at 4:53 AM Konstantin M. Khankin <
>> khankin.konstantin at gmail.com> wrote:
>>
>>> Thank you, that obviously helped. Still would be useful to know the root
>>> cause.
>>>
>>> OTOH, if sqlite creation functionality is not that critical - why would
>>> we ever enable something which only consumes time during sync for no
>>> benefit?
>>>
>>> вт, 1 сент. 2020 г. в 00:26, Dennis Kliban <dkliban at redhat.com>:
>>>
>>>> Looks like everything is up to date. I have no idea what the root cause
>>>> is, but according to this comment[0], you can work around the problem by
>>>> disabling the generation of sqlite db. I am not sure what the exact effect
>>>> of this will be for the clients that are consuming this content, but the
>>>> repository will be usable.
>>>>
>>>> [0] https://pulp.plan.io/issues/2019#note-2
>>>>
>>>> On Mon, Aug 31, 2020 at 4:10 PM Konstantin M. Khankin <
>>>> khankin.konstantin at gmail.com> wrote:
>>>>
>>>>> # rpm -qa | grep pulp | sort
>>>>> pulp-admin-client-2.21.3-1.el7.noarch
>>>>> pulp-agent-2.21.3-1.el7.noarch
>>>>> pulp-consumer-client-2.21.3-1.el7.noarch
>>>>> pulp-deb-admin-extensions-1.10.1-1.el7.noarch
>>>>> pulp-deb-plugins-1.10.1-1.el7.noarch
>>>>> pulp-docker-admin-extensions-3.2.6-1.el7.noarch
>>>>> pulp-docker-plugins-3.2.6-1.el7.noarch
>>>>> pulp-puppet-consumer-extensions-2.21.3-1.el7.noarch
>>>>> pulp-puppet-handlers-2.21.3-1.el7.noarch
>>>>> pulp-python-admin-extensions-2.0.4-1.el7.noarch
>>>>> pulp-python-plugins-2.0.4-1.el7.noarch
>>>>> pulp-rpm-admin-extensions-2.21.3-1.el7.noarch
>>>>> pulp-rpm-consumer-extensions-2.21.3-1.el7.noarch
>>>>> pulp-rpm-handlers-2.21.3-1.el7.noarch
>>>>> pulp-rpm-plugins-2.21.3-1.el7.noarch
>>>>> pulp-rpm-yumplugins-2.21.3-1.el7.noarch
>>>>> pulp-selinux-2.21.3-1.el7.noarch
>>>>> pulp-server-2.21.3-1.el7.noarch
>>>>> python-pulp-agent-lib-2.21.3-1.el7.noarch
>>>>> python-pulp-bindings-2.21.3-1.el7.noarch
>>>>> python-pulp-client-lib-2.21.3-1.el7.noarch
>>>>> python-pulp-common-2.21.3-1.el7.noarch
>>>>> python-pulp-deb-common-1.10.1-1.el7.noarch
>>>>> python-pulp-docker-common-3.2.6-1.el7.noarch
>>>>> python-pulp-oid_validation-2.21.3-1.el7.noarch
>>>>> python-pulp-puppet-common-2.21.3-1.el7.noarch
>>>>> python-pulp-python-common-2.0.4-1.el7.noarch
>>>>> python-pulp-repoauth-2.21.3-1.el7.noarch
>>>>> python-pulp-rpm-common-2.21.3-1.el7.noarch
>>>>>
>>>>> # rpm -qa | grep createrepo
>>>>> createrepo-0.9.9-28.el7.noarch
>>>>> createrepo_c-0.10.0-20.el7.x86_64
>>>>> createrepo_c-libs-0.10.0-20.el7.x86_64
>>>>>
>>>>> # cat /etc/redhat-release
>>>>> CentOS Linux release 7.8.2003 (Core)
>>>>>
>>>>> пн, 31 авг. 2020 г. в 22:48, Dennis Kliban <dkliban at redhat.com>:
>>>>>
>>>>>> This looks exactly like the issue that was reported here[0].
>>>>>>
>>>>>> What version of pulp are you using? What version of createrepo_c is
>>>>>> installed?
>>>>>>
>>>>>> [0] https://pulp.plan.io/issues/2019
>>>>>>
>>>>>> On Mon, Aug 31, 2020 at 3:16 PM Konstantin M. Khankin <
>>>>>> khankin.konstantin at gmail.com> wrote:
>>>>>>
>>>>>>> Hi!
>>>>>>>
>>>>>>> The issue still persists. Could someone take a look please?
>>>>>>>
>>>>>>> Thanks!
>>>>>>>
>>>>>>> ср, 19 авг. 2020 г., 13:15 Konstantin M. Khankin <
>>>>>>> khankin.konstantin at gmail.com>:
>>>>>>>
>>>>>>>> Hi!
>>>>>>>>
>>>>>>>> I found that my pulp2-managed mirror of
>>>>>>>> http://mirror.yandex.ru/centos/7/updates/x86_64 has not been
>>>>>>>> successfully published since April 30. I ran publish manually and received
>>>>>>>> an error:
>>>>>>>>
>>>>>>>> '''
>>>>>>>> Generating sqlite files
>>>>>>>> [/]
>>>>>>>> ... failed
>>>>>>>> Error occurred during 'sqliterepo_c' execution: Preparing sqlite
>>>>>>>> DBs
>>>>>>>>
>>>>>>>> ::
>>>>>>>> C_CREATEREPOLIB: Critical: cr_xml_parser_generic: parsing error
>>>>>>>> '/var/cache/pulp/reserved_resource_worker-1 at hive.gsk.loc
>>>>>>>> /12257118-33e7-4294-ad68
>>>>>>>>
>>>>>>>> -98e0515f2627/repodata/9e958c09c7880d130ef3321332000f43a4e4e701e416801a0f54313f5
>>>>>>>> 766d03b-filelists.xml.gz': not well-formed (invalid token)
>>>>>>>> Parse error
>>>>>>>> '/var/cache/pulp/reserved_resource_worker-1 at hive.gsk.loc
>>>>>>>> /12257118-33e7-4294-ad68
>>>>>>>>
>>>>>>>> -98e0515f2627/repodata/9e958c09c7880d130ef3321332000f43a4e4e701e416801a0f54313f5
>>>>>>>> 766d03b-filelists.xml.gz' at line: 1884123 (not well-formed
>>>>>>>> (invalid token))
>>>>>>>> '''
>>>>>>>>
>>>>>>>> I can't open .sqlite files in /tmp from CLI either:
>>>>>>>>
>>>>>>>> '''
>>>>>>>> -rw-------. 1 apache apache  54066176 Aug 19 15:12
>>>>>>>> filelists.211JP0.sqlite
>>>>>>>> -rw-------. 1 apache apache 157097984 Aug 19 15:12
>>>>>>>> primary.PQ3JP0.sqlite
>>>>>>>> -rw-------. 1 apache apache         0 Aug 19 15:12
>>>>>>>> other.X201JP0.sqlite
>>>>>>>>
>>>>>>>> [root at hive tmp]# sqlite3 primary.PQ3JP0.sqlite
>>>>>>>> SQLite version 3.7.17 2013-05-20 00:56:22
>>>>>>>> Enter ".help" for instructions
>>>>>>>> Enter SQL statements terminated with a ";"
>>>>>>>> sqlite> .databases
>>>>>>>> Error: file is encrypted or is not a database
>>>>>>>> '''
>>>>>>>>
>>>>>>>> I have multiple repos managed by pulp, some of which originate also
>>>>>>>> from mirror.yandex.ru, and they are synced and published normally.
>>>>>>>> --force-full doesn't help.
>>>>>>>>
>>>>>>>> What issue could be here?
>>>>>>>>
>>>>>>>> Thanks!
>>>>>>>>
>>>>>>>> --
>>>>>>>> Konstantin Khankin
>>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Pulp-list mailing list
>>>>>>> Pulp-list at redhat.com
>>>>>>> https://www.redhat.com/mailman/listinfo/pulp-list
>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> Ханкин Константин
>>>>>
>>>>
>>>
>>> --
>>> Ханкин Константин
>>>
>>
>
> --
> Ханкин Константин
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/pulp-list/attachments/20201002/6cc9e694/attachment.htm>