Thank you Mike, and thanks Geert! This certainly gives us some options and some things to consider.
-----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Michael Blakeley Sent: Monday, January 13, 2014 1:06 PM To: MarkLogic Developer Discussion Subject: Re: [MarkLogic Dev General] Backing up MarkLogic In general it is a bad idea to do anything with the database's filesystem. Rolling your own backup system is dangerous because any updates during the backup window will almost certainly corrupt the backup. When this happens the corrupted backup contains some files from timestamp A and other files from timestamp B, C, etc. You won't find out about this until you try to restore, because the filesystem doesn't reveal the database state. So the backup will fail just when you need it. The built-in online backup feature knows about the database state, so it can create a consistent, reliable backup at a single timestamp. Without any special knowledge about your application, you should use a conventional database backup. However there are some special cases where filesystem can work. The most basic is to shut down the cluster before taking the backup. With the database halted there are no updates, and therefore the filesystem state is consistent. Extending this idea, you can leave MarkLogic running, but at the same time ensure that there are no updates during the backup window. The best way to do this is with the forest update state, as described at http://docs.marklogic.com/guide/admin/forests#id_72520 in the docs. For example you could put all the database forests into flash-backup mode, take the filesystem backup, then resume updates. Note that any update requests made during the backup window will queue and retry. As the name implies, this is designed to work with very fast filesystem snapshot mechanisms, so that the delayed updates do not have to wait long. Also note that flash-backup adds a certain amount of complexity to the backup system. In most cases it is best to use a conventional backup. -- Mike On 13 Jan 2014, at 07:13 , Dunlap, Zachariah <[email protected]> wrote: > We are considering simply backing up the "\Program Files\MarkLogic" folder on > each machine in our cluster using Amazon Glacier instead of making backups > using the "Backup & Restore" feature. > > We'd love to get some feedback from others to try and helps us weigh the pros > and cons of simply backing up the data on the disk, vs. using MarkLogic's > backup feature. > > Thank you all in advance for your thoughts. > > -Zach > > > The information contained in this communication is intended for the > use of the designated recipients named above. If the reader of this > communication is not the intended recipient, you are hereby notified > that you have received this communication in error, and that any > review, dissemination, distribution or copying of this communication > is strictly prohibited. If you have received this communication in > error, please notify The Associated Press immediately by telephone at > +1-212-621-1898 and delete this email. Thank you. > [IP_US_DISC] > > > msk dccc60c6d2c3a6438f0cf467d9a4938 > > _______________________________________________ > General mailing list > [email protected] > http://developer.marklogic.com/mailman/listinfo/general _______________________________________________ General mailing list [email protected] http://developer.marklogic.com/mailman/listinfo/general _______________________________________________ General mailing list [email protected] http://developer.marklogic.com/mailman/listinfo/general
