mwarren@rws1.ma30.bull.com (Mark Warren) (06/13/91)
I am running nntp 1.5.10 on a Mips M2000 under RISC/os 4.51. We have one
particular nntp client that periodically becomes a tremendous CPU hog. A
typical line from 'ps -f' looks like:
UID PID PPID C STIME TTY TIME COMMAND
news 5504 32217 6 04:27:28 ? 178:33 nntpd
Note the TIME column. This was all racked up from the following rather
innocuous request (from the nntpd log file):
Jun 10 04:27:29 localhost nntpd[5504]: kelvin.uk22.bull.com connect
Jun 10 04:27:31 localhost nntpd[5504]: kelvin.uk22.bull.com newnews comp.sys.acorn,comp.sources.acorn,comp.binaries.acorn 910610 081346 local none
Here are some additional facts:
- The same client makes the same newnews request very often, and only
rarely does the situation above occur.
- The nntp client is a PC, running identical software with two other
PC's that do not cause this problem.
- The PC's are seperated from the nntp server by the Atlantic ocean,
so it is reasonable to assume long TCP turnaround time, and probably
periodic interruptions.
I have no idea why this is happening, nor can I tell where in nntpd the time
is being spent. Has anyone seen anything like this? Any hints for debugging
it? --HELP !!!--
--
== Mark Warren Bull HN Information Systems Inc. ==
== (508) 294-3171 (FAX 294-3020) 300 Concord Road MS836A ==
== M.Warren@bull.com Billerica, MA 01821 ==