Re: implementing folder hashing

2020-09-23 Thread Adrian Minta
I need to move from one mount point to multiple mount points and I'm searching for the best strategy. Right now my take is to script a rsync of the folders that need moving, change dovecot config and restart. In order to do this I need to know in advance what users will end up on the newly a

Re: implementing folder hashing

2020-09-23 Thread Manuel Delgado
I think I don't understand the question. Dovecot knows where the user's home is, you don't need to guess. With mail_home or with Passdb variables you can define the user's home location and you can use 'doveadm user' to show where the home directory is in the filesystem. However, if you are migra

Re: implementing folder hashing

2020-09-23 Thread Adrian Minta
Thank you for your reply. One questionthough, if you add a mount point how do you know what user to move ? On 9/23/20 7:35 PM, Manuel Delgado wrote: Hi Adrian, It depends much on your site configuration. In our case, we have several mount points and we distribute our user mailboxes using

Re: implementing folder hashing

2020-09-23 Thread Manuel Delgado
Hi Adrian, It depends much on your site configuration. In our case, we have several mount points and we distribute our user mailboxes using username hash (%N)[1] and limit this hash to the number of mount points. We preferred %N over %H because the distribution of users was much better. Ex. xsto

implementing folder hashing

2020-09-21 Thread Adrian Minta
Hi everyone, I want to ask if any of you has done a migration from a single storage to multiple back-end storages and what is the best strategy for folder hashing that you use. Using something like %H or %M how do you predict where the mail will end up in order to move the existing folders ?