1.3.70 NNTP server uptime good

Kevin Sawyer (sawyerk@apci.net)
12 Mar 1996 06:12:55 GMT


Our NNTP server has been up for 6 days and 10 hours running kernel 1.3.70.
I think that this is particularly significant due to the fact that we have
as many as 100 readers at any given time, 3 incoming feeds, and three
outgoing. We used to have SCSI and net errors all the time, but no more
since we upgraded to a DPT PCI caching/RAID controller (with 4 4GB drives
striped) and an SMC PCI ethernet card.

We used to have only an Adaptec 1542CF (ISA) with 1 540MB drive and 1 9GB
drive for our spool, and an AMD1500-based LANCE ethernet adapter. The
network seemed rock solid, but we were constantly getting SCSI errors such
as "run out of queue slots" and the box would just lock up. So, we trashed
the 9GB and went to our current 16GB array. The 1.2.13 source woudn't
support our DPT for some reason, so I thought "what the hell" and compiled
1.3.70. The drives came right up, but so did hunreds of "socket destroy
delayed" messages on the console (each minute). So, I plucked out the LANCE
card and tossed in the SMC PCI and recompiled. Voila! Now I only have 3-4
"socket destroy delayed" messages per day.

Other pertinent data (maybe):

American Digicom main board (I think)
256k write-back cache
64MB 60ns parity RAM
Intel Pentium 90Mhz
Some cheesy ISA VGA card (Trident, I think)
INN 1.4sec

For what it's worth...

Kevin

--
 --------------------------------------------------------------------------
 Applied Personal Computing, Inc.  Home of the APCi MultiPlayer Game Server 
 Voice: 618.632.7282 / 800.535.APCi   FAX: 618.632.7287   BBS: 618.628.APCi
 telnet://bbs.apci.net        ftp://ftp.apci.net        http://www.apci.net
 --------------------------------------------------------------------------