The servers with "blade" in their names all seem to be running 3.1.1 (since about March 20th 2006), but I can confirm that two of the new boxes, "filter7" and "filter8" are running 3.1.0. However, I looked at some items that got through to my inbox overnight and on almost identical spam items, the "filter" servers actually had given a higher score than blade1 did, which is a bit counter-intuitive.
So, I looked into my Held Mail and did a little more analysis. I have six similar "Investment Strategy" spams (all identical subjects and almost identical spam bodies, except for the URLs). Here are the receipt times, server names, and SA analyses from these items:
CODE
Received: (qmail 29336 invoked from network); 1 Nov 2006 17:05:28 -0000
X-Spam-Checker-Version: SpamAssassin 3.1.1 (2006-03-10) on blade6
X-Spam-Level: *********************
X-Spam-Status: hits=21.6 tests=SARE_CSNUMTAG,SARE_RMML_Stock4,
UNPARSEABLE_RELAY,URIBL_AB_SURBL,URIBL_BLACK,URIBL_JP_SURBL,
URIBL_OB_SURBL,URIBL_SBL,URIBL_SC_SURBL,URIBL_WS_SURBL version=3.1.1
Received: (qmail 2814 invoked from network); 1 Nov 2006 20:16:37 -0000
X-Spam-Checker-Version: SpamAssassin 3.1.0 (2005-09-13) on filter8
X-Spam-Level: *****************
X-Spam-Status: hits=17.5 tests=SARE_CSNUMTAG,SARE_RMML_Stock4,
UNPARSEABLE_RELAY,URIBL_AB_SURBL,URIBL_JP_SURBL,URIBL_OB_SURBL,
URIBL_SBL,URIBL_SC_SURBL,URIBL_WS_SURBL version=3.1.0
Received: (qmail 7781 invoked from network); 2 Nov 2006 10:19:06 -0000
X-Spam-Checker-Version: SpamAssassin 3.1.1 (2006-03-10) on blade5
X-Spam-Level: *********************
X-Spam-Status: hits=21.6 tests=SARE_CSNUMTAG,SARE_RMML_Stock4,
UNPARSEABLE_RELAY,URIBL_AB_SURBL,URIBL_BLACK,URIBL_JP_SURBL,
URIBL_OB_SURBL,URIBL_SBL,URIBL_SC_SURBL,URIBL_WS_SURBL version=3.1.1
Received: (qmail 14712 invoked from network); 2 Nov 2006 10:18:09 -0000
X-Spam-Checker-Version: SpamAssassin 3.1.1 (2006-03-10) on blade1
X-Spam-Level: *********************
X-Spam-Status: hits=21.6 tests=SARE_CSNUMTAG,SARE_RMML_Stock4,
UNPARSEABLE_RELAY,URIBL_AB_SURBL,URIBL_BLACK,URIBL_JP_SURBL,
URIBL_OB_SURBL,URIBL_SBL,URIBL_SC_SURBL,URIBL_WS_SURBL version=3.1.1
Received: (qmail 15251 invoked from network); 2 Nov 2006 10:19:16 -0000
X-Spam-Checker-Version: SpamAssassin 3.1.1 (2006-03-10) on blade1
X-Spam-Level: *********************
X-Spam-Status: hits=21.6 tests=SARE_CSNUMTAG,SARE_RMML_Stock4,
UNPARSEABLE_RELAY,URIBL_AB_SURBL,URIBL_BLACK,URIBL_JP_SURBL,
URIBL_OB_SURBL,URIBL_SBL,URIBL_SC_SURBL,URIBL_WS_SURBL version=3.1.1
Received: (qmail 21148 invoked from network); 2 Nov 2006 11:14:12 -0000
X-Spam-Checker-Version: SpamAssassin 3.1.0 (2005-09-13) on filter7
X-Spam-Level: ****************
X-Spam-Status: hits=16.2 tests=SARE_RMML_Stock4,UNPARSEABLE_RELAY,
URIBL_AB_SURBL,URIBL_JP_SURBL,URIBL_OB_SURBL,URIBL_SBL,URIBL_SC_SURBL,
URIBL_WS_SURBL version=3.1.0
There does seem to be a bit of inconsistency, and the "blade" servers came out ahead, specifically because the "filter" servers didn't seem to know about the "URIBL_BLACK" test that the blade servers identified. The only other difference was the the last one, processed by "filter7" is missing the "SARE_CSNUMTAG" flag, but it's possible that there was something different about that item.
So, I'd say this small, but carefully selected sample, might indicate that there's some work to be done to make the filtering on all of the servers equal to one another, but I'm speaking as a person who has never installed or configured SpamAssassin. :-)
Wazoo....wanna kick this to JT also, or should I send him an email?
DT