Hi all,

We are testing and validating a crash/restore situation with Courier 0.35.1 compile for Solaris 8 with GCC 2.95.3 package from SunFreeware.

We experience some problems at Courier startup (after restoring the entire Courier workspace) with some mails allready in queues.

Before saving Courier, we send some messages and stop Courier before those was sent, to check a save/restore with some mail in queue of Courier. We save all files/directory, and finally we restore all (on the same machine). Before restoring we destroy all files and directory owned by Courier.

We use NetBackup ( a commercial product ) to accomplish the "save/restore" step. All FS under our machine are UFS. I think the save process is made by a TAR clone, not a cpio one, neither a "DD".

At re-startup, Courier send a lot of :

Courierd : Queue file corruption inode 57061 Courierd : Queue file corruption inode ... Courierd : Queue file corruption inode ...

in syslog.

Those inodes point on files in queues dir.

Does anyone have any idea of this problem ? In fact, does anyone have the "best way" to make "a perfect crash/restore procedure" of Courier ?

Thanks in advance for your respons and your time.

Best regards