On 3/31/2014 1:25 PM, Viktor Dukhovni wrote:
On Mon, Mar 31, 2014 at 05:14:49PM +0000, Viktor Dukhovni wrote:However the particular fatal log message you report "open database: ..." only occurs in one place in Postfix: dict_sdbm.c: msg_fatal("open database %s: %m", dbm_path); You must be one of the folks who never got the memo about not using "sdbm". :-) Switch to Berkeley DB "btree" for your scache databases.Or perhaps not, my RTFS search was too narrow to find the other cases. dict_cdb.c: "open database %s: %m", cdb_path)); dict_db.c: "open database %s: %m", db_path)); dict_dbm.c: "open database %s: %m", dict_lmdb.c: "open database %s: %s", mdb_path, mdb_strerror(status)); dict_sdbm.c: msg_fatal("open database %s: %m", dbm_path); dict_thash.c: "open database %s: %m", path));
It looks like running out of space causes that too. I found this entry in the logs from the day before:
fatal: /var/lib/postfix/smtpd_scache.db: flush dictionary: No space left on device
One of the first things I checked but space was fine when i looked. It looks like a backup was using the drive as temporary space and then it would move the backup file to smb share when it was finished. So while it was using it as temporary storage it would fill up the drive, thus the problem. It has since been fixed.
Thanks for your help
