One of the nntp servers that nntpcache connects to went down today, and I discovered that it causes problems. I would have thought that a server being down would just cause nntpc to return what it has in the cache, but instead the reader connection just hangs until the network timeout setting is reached and returns a bogus "this machine does not have permission to use the [local nntpc server] server" message (trn) or "Message from iserver is: 502 NNTPcache master server unreachable (knews). This also prevents groups from the second (still working) server to be read. Also, I don't think articles are really being cached properly. The bodies are going onto disk as they are fetched and read, but don't seem to be read from the cache afterwards (i.e. large articles are loaded into a reader at our outside ISDN feed speed instead of local ethernet speed even after being read once, Also, after the article has expired off the feed (but still in the cache), attempts to select it give the "bad article number, maybe it has expired off the server" message. The actives and overviews seem to be cached ok and can feed the readers at a fast clip (just the bandwidth saving from avoiding multiple downloads of the 330k active file make nntpc worthwhile). A recent nntpcache.stats file supports my theory: NNTPCACHE 0.88.3UL statistics on server.spec.com port 119 OVERALL: Statistics gathering commensed: 06 Jun 1996 17:54:12 CDT Current nntpcached started: 14 Jun 1996 23:53:27 CDT Servers run during stistical period: 10 Clients active: 1 Client connects: 911 Outing server connections: 877 Outing server connections (failed): 10 Total data from remote servers (bytes): 268.21M Total data to remote servers (bytes): 0b Total data from clients (bytes): 0b Total data to clients (bytes): 0b Total cache cache efficiency: -268208105.00% IPC messages to children (bytes): 6.12k IPC messages to children: 4185 IPC messages to children (bytes): 6.12k IPC messages from children: 5094 IPC messages from children (bytes): 638.18k Total time elapsed in children: 3d7h41m38s Database stores: 4178 Database fetches: 0 Database size: 511.18k "group" commands: 8243 "listgroup" commands: 20 Groups cached with > 0 articles: 122 Groups expired: 0 Articles expired: 0 Posts: 7 Posts (crossposted): 0 Posts (failed): 0 Posts (bytes): 0b Incoming crossposts: 1070 Incoming crossposts (bytes): 77.96M LISTS: List-name---- --out-of-cache -----into-cache hits --xfer-per-hour lines -length active 4147 4.92M 0 0b 100% 15.37 18.22k 1 308.05k active.times 0 0b 0 0b 100% 0.00 0b 0 0b newsgroups 2 86.25k 0 0b 100% 0.01 320b 586 42.87k overview_fmt 0 0b 0 0b 100% 0.00 0b 8 65b MESSAGES: Data-type---- --out-of-cache -----into-cache hits ----------proxy xfer-per-hour article 0 0b 2797 243.57M 0% 0 0b 10 902.55k head 0 0b 921 610.25k 0% 0 0b 3 2.26k body 0 0b 460 3.56M 0% 0 0b 2 -76b xover 465830 88.83M 37214 15.86M 85% 0 0b 1864 387.92k xhdr 0 0b 0 0b 100% 0 0b 0 0b msgid 0 0b 4178 137.88k 0% 15 511b -- Frank Smith -- System Administrator E-mail: fsmith@spec.com Systems & Processes Engineering Corp. (SPEC) Voice:(512) 306-1100 x154 401 Camp Craft Road Fax: (512) 306-1122 Austin, TX 78746-6558 Web: http://www.spec.com
