Hi folks. I continue to fight with regular locking database damage in Linux box. See Bugzilla Bug 370. The current question to gurus: what is the RIGHT behavoir of locking.tdb size? Should it decrease when the number of records decreases? Or, it should stay at maximal reached capacity? Alexey
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Alexey Lobanov wrote: | Hi folks. | | I continue to fight with regular locking database damage in | Linux box. See Bugzilla Bug 370. | | The current question to gurus: what is the RIGHT | behavoir of locking.tdb size? Should it decrease when | the number of records decreases? Or, it should stay at | maximal reached capacity? we never pack TDB's. You can safely remove the tdb after shutdown though. What size are you seeing? cheers, jerry - ---------------------------------------------------------------------- ~ Hewlett-Packard ------------------------- http://www.hp.com ~ SAMBA Team ---------------------- http://www.samba.org ~ GnuPG Key ---- http://www.plainjoe.org/gpg_public.asc ~ "You can never go home again, Oatman, but I guess you can shop there." ~ --John Cusack - "Grosse Point Blank" (1997) -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQE/ex/SIR7qMdg1EfYRAkP1AJ0Xb5yi0yH4r8ajZbhzHNgObty50ACgm9+B zf/5syr5SYvt3X6kEQj71eA=Rgdf -----END PGP SIGNATURE-----
Hi Gerald. Date sent: Wed, 01 Oct 2003 13:41:22 -0500 From: "Gerald (Jerry) Carter" <jerry@samba.org> To: Alexey Lobanov <aal@evidence-cpr.com> Copies to: samba@lists.samba.org Subject: Re: [Samba] Size of locking.tdb?> we never pack TDB's. You can safely remove the > tdb after shutdown though. What size are you seeing?Great. I know wtf. Linux 2.4.any, Samba 2.2.any. When TDB's reside on "async" mounted ext2 partition, locking.tdb size FOLLOWS the number of records. From 25 kilobytes to hundreeds of kilobytes, and back. When /var is mounted "sync", the database really stays at maximum size (now 548864 bytes after >3000 records at peak). "use mmap" does not affect this behavior. It is difficult for me to say where is the exact bug resides: in Samba or in Linux 2.4 kernel. But the name of bug seems to be "non-coherent access". It is also difficult to say now if "sync" mounting is a sufficient workaround. At least, it is not fatal for performance if /var/spool and /var/log are mounted separately. And it is funny that this TDB "packing" which definitely happens at async Linux filesystem is not necessary fatal for locking database contents. Looks like in this case Samba dies in quite rare set of conditions. Maybe, when the number of records falls very quickly from 2000 to 50, "accounting team go home". Maybe not. Thank you! Alexey