[Spacewalk-list] SpaceWalk High Availability

[Nelson Marques]

I'm not really sure what I'm going to do for now; Going to wait until
monday til my colleague arrives and eventually something like this:

 * Upgrade to PostgreSQL 9.2 (synchronous replication);
 * Use DRDB for the storage holding the repositories (http://www.drbd.org/),
we have 2Gbit between our datacenters, so should be ok with that
 * One machine in each datacenter and get them working with
heartbeat/pacemaker

Seems probably a quick way to go. I'll let you know.

NM


2014-02-28 16:08 GMT+00:00 Justin Edmands <shockwavecs at gmail.com>:

> On Fri, Feb 28, 2014 at 10:32 AM, Nelson Marques <nmo.marques at gmail.com>wrote:
>
>> Hi Justin,
>>
>> Thanks for replying; Regarding Backups I'm doing backups with bacula on
>> this items:
>>
>>  * PostgreSQL DB (filesystem);
>>  * /etc
>>  * /var/lib/tomcat6 and /usr/share/tomcat6 (redundant)
>>
>> The DB backup is made like this:
>>
>>  * (pre-backup) psql -c "SELECT pg_start_backup('bacula-backup-${DATE}',
>> true);"
>>  * backup stuff (/var/lib/pgsql/data)
>>  * (post) psql -c "SELECT pg_stop_backup();"
>>  * (post) VACUUM ALL
>>  * (post) reindex --all
>>
>> So I have that covered also; The thing is, HA was required so that we
>> don't have complains from the devs having their orchestration stuff broken
>> (ansible) due to spacewalk unavailability. Recovering backups takes it's
>> time.
>>
>> Aditionally I also have backups done on HP DataProtector from the whole
>> VM. Either way, I have to go HA despite of the backup level. I was just
>> trying to figure what people are running and what kind of setup would
>> apply, eventually balancing stuff, hot standby or any master/slave
>> configuration. How it can be split appart (ex: DB being replicated, shared
>> NFS mount for package repo's and frontends).
>>
>> NM
>>
>>
>>
>>
>> 2014-02-28 14:23 GMT+00:00 Justin Edmands <shockwavecs at gmail.com>:
>>
>>
>>> > On Feb 28, 2014, at 7:08 AM, Nelson Marques <nmo.marques at gmail.com>
>>> wrote:
>>> >
>>> > Hi all,
>>> >
>>> > Currently at work we use spacewalk to manage a small part of our
>>> infra-structure (around 100 machines running centos 6 from 3 environments).
>>> Recently there was a problem with our SAN storage ending up with the
>>> filesystem on the spacewalk machine having a few bad sectors (postgresql db
>>> got corrupt as consequence).
>>> >
>>> > Our development team has spacewalk integrated with ansible for
>>> orchestration; If for some reason (as above) spacewalk is unavailable
>>> ansible seems to break also there is no way to install software on machines.
>>> >
>>> > I'm currently looking for documentation or some 'enlightenment' on the
>>> best procedures to set spacewalk for high availability or a recommended
>>> architecture so that we have redundancy for it. Anyone can point up some
>>> docs or share experiences?
>>> >
>>> >
>>> > NM
>>> > _______________________________________________
>>> > Spacewalk-list mailing list
>>> > Spacewalk-list at redhat.com
>>> > https://www.redhat.com/mailman/listinfo/spacewalk-list
>>>
>>> I also have a small environment of about 100 machines and just do
>>> nightly Postgres backups. The entire machine, a VM, is backed up to virtual
>>> tape. Given it being a small setup, HA would be far too involved. No
>>> database backups I'm assuming?
>>>
>>> _______________________________________________
>>> Spacewalk-list mailing list
>>> Spacewalk-list at redhat.com
>>> https://www.redhat.com/mailman/listinfo/spacewalk-list
>>>
>>
>>
>> _______________________________________________
>> Spacewalk-list mailing list
>> Spacewalk-list at redhat.com
>> https://www.redhat.com/mailman/listinfo/spacewalk-list
>>
>
>
> Oh nice, we use bacula as well. Great to hear you have a solid backup
> solution. I have no further content to add. I remember looking into the
> same kind of setup a while back and ultimately wound up on satellite
> documents. An example would be this one:
> https://gb.redhat.com/f/pdf/rhn/Satellite-HA.pdf
>
> I would attempt to build the RHCS on two VMs and then import the existing
> database. LEt us know how it goes.
>
> _______________________________________________
> Spacewalk-list mailing list
> Spacewalk-list at redhat.com
> https://www.redhat.com/mailman/listinfo/spacewalk-list
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20140228/ee0062f7/attachment.htm>