couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Randall Leeds <randall.le...@gmail.com>
Subject Re: [ANN] Unofficial RPMs for RHEL-5/CentOS-5 and derivatives
Date Wed, 13 Oct 2010 17:15:20 GMT
I know, Noah, but unless the setup uses the heartbeat option the new Couch
might start before the old one goes away, crash with eaddrinuse, and without
any error from the init script's restart() the user finds themselves without
a running couch.

I don't know why, but I've only had this happen to me on CentOS.
On Oct 13, 2010 6:55 AM, "Noah Slater" <nslater@apache.org> wrote:
>
> On 12 Oct 2010, at 20:55, Randall Leeds wrote:
>
>> 1) Based on what I saw in thy MySQL package, stop() asks CouchDB to
>> quit and then checks every second (up to 60 seconds) to see if it's
>> shut down before forcibly killing the process. This avoids an issue
>> where restart could sometimes crash the new CouchDB process with
>> EADDRINUSE when it tries to bind the server socket by making sure that
>> the old CouchDB process is actually stopped before starting the new
>> one.
>
> Why bother? CouchDB is designed to be kill -9'd. :)

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message