Fixed: OpLocks caused the corruptions/slowness (Was: How Samba let us down)

[Volker.Lendecke at SerNet.DE]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> Clients commonly ignore oplock breaks because of network
> problems (borderline hubs etc.). Many people are suffering
> from network hardware that performs adequately in light
> use situations and fails under heavy load. I myself have
> ended up junking hubs with this problem.

Just my 2 cents:

I recently had a case where the log.smbd looked like this. Oplock
break failures. The customer told me again and again that he does not
have any hardware problem. So I went to the site, and he was right. It
was a problem with nss_ldap. One smbd opened a file with oplocks. It
then had to do something different, calling nss_ldap. nss_ldap then
went to sleep, ignoring *all* signals except -9. nss_ldap would not
come back. So the oplock break request never made it to the smbd that
held the file open.

Volker

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Key-ID D32186CF, Fingerprint available: phone +49 551 3700000

iD8DBQE9wjsjOmSXH9Mhhs8RAo2GAJ9j5tuwykka3ielhsSgh0ef0koWtACeLnzR
2j20v7sbKT9qa/IURa4pgBE=
=Ygp1
-----END PGP SIGNATURE-----