Please consider the 2 cases of: - a planned reindex, since we need to speed up the DB. - an unplanned reindex, which happens after an error or improper shutdown, etc.
I wonder if OrientDB checks the index schema during a cluster sync? If there needs to be a 100% identical schema, then we can never reindex without at least some downtime. On Friday, December 12, 2014 12:20:39 AM UTC-5, Simon L wrote: > > Hi guys- > > We have a mission-critical cluster that must be available 24x7 and take a > very, very heavy load. > > - It seems like reindexing is inevitable. Eventually, the database will > require a reindex for one reason or another (either planned or unplanned.) > - OrientDB seems to halt all DB activity when it's reindexing. > - If a DB needs reindexing, how can we keep our cluster up during that > time? Has anyone else done this in production? > > - Keep in mind that our DB gets 500 writes per second, and so removing 1 > node at a time to reindex may not won't work. It will be out of date by the > time it finishes reindexing and rejoins the cluster. > - Also, I suspect that a reindexing database will typically affect all > nodes in the cluster, causing all nodes to go down during reindexing. My > tests have shown this so far, but I've only done limited testing. > > Please let me know the status of this situation right now, and plans for > the future to deal with reindexing. > -- --- You received this message because you are subscribed to the Google Groups "OrientDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
