From: Johan Andersson Date: February 27 2012 6:32am Subject: Re: Cluster Failure Handling .. List-Archive: http://lists.mysql.com/cluster/8260 Message-Id: MIME-Version: 1.0 (Apple Message framework v1257) Content-Type: multipart/alternative; boundary="Apple-Mail=_69D6D493-2A2B-4B38-9EC8-94E2B1D8B608" --Apple-Mail=_69D6D493-2A2B-4B38-9EC8-94E2B1D8B608 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=iso-8859-1 Perhaps you can use geo rep and replicate to another Cluster.=20 Otherwise, there is no magic. If a cluster fails, it has to be restarted = and it can take time. BR johan On Feb 27, 2012, at 3:51 AM, umapathi b wrote: > Anybody there to comment/help on this please ?! >=20 > - Umapathi >=20 > On Thu, Feb 23, 2012 at 4:57 AM, umapathi b = wrote: > Hi All, >=20 > I have a production cluster running with 2 data nodes , 2 sql nodes = and 1 mgmt node . > And I have a slave to one of the above servers with innodb plugin for = data backup purpose=20 > which is running fine . >=20 > One day , while trying to do some parameter changes wrt disk based = tables , I got some error=20 > and the cluster was not able to re-start/recover . In this case , I = had to start the cluster with --initial=20 > option again and reload/restore the data from the slave . But this = took considerable time(around 2 hours) .. > and I was safe as it was off-peak time ..and did not impact the = customers. >=20 > How can I handle this kind of complete failure of cluster , in order = to have no downtime at all=20 > or to quickly recover ?! >=20 > I am sure somebody might have faced this kind of issue earlier ... = Advice/Guidance in this regard=20 > is highly appreciable .. >=20 > Thank you all in advance .. >=20 > - Umapathi. >=20 >=20 >=20 >=20 >=20 >=20 >=20 --Apple-Mail=_69D6D493-2A2B-4B38-9EC8-94E2B1D8B608--