![]() inject the 5 files into it from the hard disk (this is very fast) and in one operation. I also had a look at the node that was still recovering and saw that CPU usage was somewhat elevated (50%), but not maxed out. we get an exact copy (as with the DOS DiskCopy command). One is still pending at took another 5 minutes to be fully recovered. 4 out of 5 shards have been recovered fully. Speed throttling: Slow down backups if they are too fast. This is about 5 minutes into the recovery process. To set up a exact plain file copy task in BackupChain, click New Task, and select Universal. ![]() I had a look at the recovery endpoint and found this: Note that we only need a byte for byte copy, there's no need to modify templates, filter data. What are other options to do an exact copy of an index in the fastest way?Īlternatively, how I can troubleshoot / optimize the snapshot/restore method to get back to previous timings? We also verified that this happens regardless of which node is restoring those last shards, so it's not node-related. This option will copy all files (excluding subfolders) from C:src to C:dst. The remaining 1 or 2 shards take 10 minutes or more however to recover. The most basic use of robocopy is using a source and destination directory with no options. What is strange is that 3 or 4 out of 5 shards still recover in about 3-4 minutes. After our index has grown however, it now takes between 10 and 15 minutes.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |