Hi Bill, Thanks a lot for your detailed email, it helps a lot to understand how can we do it correctly.
I will follow your recommendations. Have a nice day :) Guille El sáb, 22 ene 2022 a las 21:07, Bill Arlofski via Bacula-users (< bacula-users@lists.sourceforge.net>) escribió: > On 1/22/22 08:08, Guillermo Martin via Bacula-users wrote: > > Hello everyone, > > > > I've setup bacula server 11.0.5.7 community edition with libs3 and > postgres in centos 7 to upload the backups in AWS S3 > > > > Everything is working fine but I have a doubt with Archive Device as we > just want to save the backups in aws s3 (one > > location), not in local disk and aws. If we setup a local server path it > saves data in local server and aws s3, if I mount > > the s3 bucket like this example, it duplicate the backups. If I > remove"Archive Device" line it doesn't work... I couldn't > > find how to do this in bacula documentation > > > > I don't know how to set up the bacula server to save the backups just in > AWS > > > > Any one could explain me please how can I do it? > > > Hello Guille, > > The S3 Cloud plugin (well, actually all of the SD cloud plugins) were > designed with a couple things in mind: > > - The cloud companies are usually happy to take in your data for free, but > they charge egress fees when you download it. > > - Keeping this in mind, the local cache was designed so that you could > have fast, local (ie: free) restores for some of your > data - data that has not been pruned or purged from the local cache. > > - Then, for other data that must be retrieved from the cloud, the cloud > volume 'parts' were designed so that you have control > over the size of each part (MaximumPartSize). This way, Bacula will be > able to download the least amount of cloud parts when > you are restoring some files from a job. (again, trying to keep your > download costs down) > > > There are a couple settings that you have control over to help keep the > required amount of local cache storage to a minimum: > > - TruncateCache = AfterUpload (Options: No, AfterUpload, AtEndOfJob - > default is No) > - Upload = EachPart > > Using these two Cloud resource settings Bacula will upload each part as > soon as it is finished writing to the local cache, > and then, once the part has been successfully uploaded to the Cloud, it > will be truncated (ie: deleted) > > You can also speed things along with these settings: > > - MaximumConcurrentUploads: The default is 3 > - MaximumConcurrentDownloads: The default is 3 > - MaximumUploadBandwidth: The default is unlimited > - MaximumDownloadBandwidth: The default is unlimited > > > And finally, consider that using the above TruncateCache and Upload > settings will keep the required space in the local cache > to a minimum during backups, if you need to restore an entire job, the > FULL space the job used will be required in the cache > because Bacula will download all the required parts to complete the > restore to the local cache, then, once all parts have > been downloaded the data will be sent to the FD. > > There is a feature request for the SD to stream the restore directly to > the FD, but I am not sure if this has made the roadmap. > > > Hope this help, > Bill > > -- > Bill Arlofski > w...@protonmail.com > > > > _______________________________________________ > Bacula-users mailing list > Bacula-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/bacula-users > --
_______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
