Re: 10-year cluster uptime anniversary
- From: Ed Wilts <ewilts@xxxxxxxxxx>
- Date: Sun, 31 May 2009 15:52:53 -0700 (PDT)
On May 30, 8:59 am, Tom Dahl <nikond...@xxxxxxxxx> wrote:
Ed, that is very cool! Thanks for sharing it.
How many nodes are in the cluster? Are they distributed over more than
The nodes are in 2 data centers, about 500m apart in our campus
(separate buildings). There are currently 3 nodes at one site and 2
in the other. Initially they were connected via FDDI using DEC
Gigaswitches. Today, they're connected via multiple GigE paths and
dual fibre connections to the SAN. They boot from EVA and all of the
storage is host mirrored between data centers.
The cluster has gone through at least one if not 2 major releases
since it was formed. It's at 7.3-2 now but I can't remember if we
were at 5.5 or 6.x when I formed it.
Its primary purpose is to run a commercial publishing application,
I remember that the Gigaswitches used to have nasty habit of silently
dropping one of their paths. And then one day, it silently dropped
both paths and we were wondering why the shadow copies were going so
slowly...they were running over a 100Mbps connection, shared with the
interactive traffic. It's a miracle that it worked at all! We reset
the FDDI paths, the shadow copies started going faster, and we added
some additional monitoring to make sure it would be detected if it
Long live VMS!
- Prev by Date: alpha_ia64.pdf may be helpful (or may not)
- Next by Date: Re: alpha_ia64.pdf may be helpful (or may not)
- Previous by thread: Re: 10-year cluster uptime anniversary
- Next by thread: Re: 10-year cluster uptime anniversary