fingerhu@ircam.fr (Michel Fingerhut) (03/04/91)
From time to time (i.e., several times a day), a command started from csh on our 5820 (DEC/RISC Ultrix 4.0) takes 15-30 seconds to start, regardless of the load average of the machine. A "ps axl" on the controling terminal shows it's actually the csh which is in disk wait, with funny flags (some digits missing): in the following output it's the last line. F UID PID PPID CP PRI NI ADDR SZ RSS WCHAN STAT TT TIME COMMAND 11008001 0 4281 222 2 1 0 2bdf 60 48 5f05c I p2 0:38 rlogind 1180c001 70 4282 4281 0 -1 0 2fd7 0 0 4314 D p2 0:00 -csh (csh) 10008021 0 5543 4282 0 25 0 2d32 232 188 T p2 0:04 -u (csh) 9001 70 10143 4282 1 -5 0 3030 368 284 c224 D p2 0:09 -csh (csh) Any clues, help? It is rather annoying to have to wait 30 secs for an ls of a directory on the local disk containing 4 files on a machine with load avg of 1.5 and presumed to work at 35 MIPS.
pavlov@canisius.UUCP (Greg Pavlov) (03/06/91)
In article <1991Mar4.125200.10446@ircam.fr>, fingerhu@ircam.fr (Michel Fingerhut) writes: > From time to time (i.e., several times a day), a command started from csh > on our 5820 (DEC/RISC Ultrix 4.0) takes 15-30 seconds to start, regardless of > the load average of the machine. It was my impression that what you are seeing is the symptom of problems in ULTRIX multiprocessing for which there is supposed to be at least partial relief in version 4.1. But I may be wrong; we no longer have our 5810 so I lost interest in following its vagaries.. pavlov@stewart.fstrf.org