I recall an issue with SCOS if the common store db was corrupt, then mailusers wouldbe interrupted.
With the centralization of attachments via DAOS wuouldn’t there be a similar problem while the central db had to be resoted from disk?
Otherwise, I like the idea
tx
Subject: domino may keep running but atts would be lost
Daos is composed of the DAOS directory and some configuration files, and a database.If the configuration files or database is corrupted, then atts would be lost.
So Domino should keep running but all users using DAOS will loose their atts.
In this case you should restore a backup and cross your fingers.
This is the usual problem where you have centralized things… the code from SCOS is completely different, but the idea of a central repository is the same… and the risks are the same.
That’s why I don’t encourage DAOS when I am asked, because on one side it saves space (only if the atts are repeated) but on the other side administrators will have more things to mantain (TS, clusters, a dedicated Domino backup since a generic backup utility does not work with DAOS, file system issues since there will be billions of files - imagine a chkdisk if server fails).
Subject: DAOS and object store corruption
DAOS requires transaction logging to be enabled on the DAOS object store server. If you are properly backing up the server and there is corruption at some point you would be able to recover using a tape restore.
Subject: Thanks Mr Oso
I am well familiar with TL, but does that mean there is no interruption of Domino service?
ANother words, can the Domino server stay up during recovery?
Subject: I think so…
You can restore eg. mailfiles using TL with the server up.They claim DAOS is a completely new concept, and no long-distance relative to Shared Mail.
Let´s hope it´s so.
Exiting stuff…
Subject: The difference
between restoring a mail file and the central attachment store is the latter is shared.
From the other response perhaps a clustered failover server would work.
Subject: DAOS high availability
To insure the highest availability your best bet would be to have a clustered failover server with DAOS enabled.