Hi all, update:
it seems that sometimes one node cannot catch up with the journal. It slows down to about one revision per second. Has somebody any idea why this could happen? Further observation shows, that if this "slow" node tries to write an item this is not possible, it takes forever. If another node from the cluster then tries to write to the same item it is also blocked. Could this sort of deadlock really happen? What is going on behind the scenes? Any hints? were stuck... regards, markus Am 27.09.2010 13:12, schrieb Markus Blaurock: > Hi all, > > we are experiencing strange behaviour in our cluster of 8 > jackrabbit-nodes (with sling, jr version 1.6.1). > > Sometimes each node in our cluster has a write-blockade. It is > impossible to make a simple session.save(), even writing a new node is > not possible. > > This block stays until we shut down a specific node. Which one is > responsible varies, e.g. we don't know in adavance which node to > shutdown, until we did it. > > We would like to trace down the error, so could someone give us a hint > where to look? > What coincidence could be responsible for preventing a whole cluster of > writing to a repository? > > regards, > markus > > > > mit freundlichen Grüßen Markus Blaurock -- DIG GmbH Neckarstraße 1/5, 78727 Oberndorf am Neckar, Fon: +49 7423 8750 62 Registergericht: Amtsgericht Stuttgart HRB 480914 Geschäftsführer: Carsten Huber
