On Wed, 22 Jun 2016, Joe Blow wrote:
I'm using pacemaker and corosync for drbd/gfs2/ILO STONITH management and
keepalived just for vrrp.
I'd just typically used keepalived for VRRP/VIP stuff so I reached for that.
Ok, I default to using pacemaker/corosync (from using heartbeat before that),
and it's trivial to have it manage a VIP once you're running it.
As far as drbd, all of the boxes doing that are connected via crossover cable
and are running dual-primary. I'd be interested in the replication across
multiple DCs. That's a much more labourous task.
One thing that does happen is that each system writes to /storage(gfs2
clustered filesystem via drbd)/$hostname. So both systems get a copy of each
other under a different folder and they never write to each other's directory.
You have to grep through 2 folders but since the dir structure is identical,
it's just another wildcard.
I've thought about doing that, but it's just been too many complex pieces for me
to be happy with it. So I tend towards the path of two systems receiving all
logs and then declare one the 'gold copy' after the fact.
With CLUSTERIP (documented in my usinix presentation) and UDP logs, I can have
both systems in one DC receving identical log streams. the resulting files
won't quite be identical due to race conditions in rsyslog and exact timing of
log rotation, but for practical purposes they will be the same. By then having
one copied to the other based on which is the 'primary' I've got a rsyncable
master 99%+ of the time, and the remainder of the time I have the individual
files that I can inspect and merge manually if it absolutly matters (it seldom
does so I just pick one and use it)
David Lang
Cheers,
JB
Original Message
From:[email protected]
Sent:June 22, 2016 8:41 PM
To:[email protected]
Reply-to:[email protected]
Subject:Re: [rsyslog] Central logging solution
On Wed, 22 Jun 2016, Joe Blow wrote:
If you're trying for true HA/prod setup, I'd suggest looking at
DRBD+pacemaker+corosync+crmsh+keepalived+rsyslog with local storage.
Why would use keepalived as well as pacemaker/corosync? It seems to me that you
would use one or the other.
I've given serious consideration to setting up pacemaker/corosync between my two
archive servers (in two different datacenters) and having the 'active' one copy
data to the 'inactive' one as the logs roll, just so that the arcives are
identical so that rsync can patch gaps (and when it can't deliver the messages
to the remote system, keep both copies as they will be different)
I haven't completely satisfied myself that I cover all failure modes with this,
so I haven't implemented it yet.
David Lang
_______________________________________________
rsyslog mailing list
http://lists.adiscon.net/mailman/listinfo/rsyslog
http://www.rsyslog.com/professional-services/
What's up with rsyslog? Follow https://twitter.com/rgerhards
NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of
sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE
THAT.
_______________________________________________
rsyslog mailing list
http://lists.adiscon.net/mailman/listinfo/rsyslog
http://www.rsyslog.com/professional-services/
What's up with rsyslog? Follow https://twitter.com/rgerhards
NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of
sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE
THAT.
_______________________________________________
rsyslog mailing list
http://lists.adiscon.net/mailman/listinfo/rsyslog
http://www.rsyslog.com/professional-services/
What's up with rsyslog? Follow https://twitter.com/rgerhards
NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of
sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE
THAT.
