krowitz@RICHTER.MIT.EDU (David Krowitz) (10/30/90)
Well, I'm trying to building the X11 R4 sources on our DN3000's, 3500's, and 2500's without much luck. What happens is that the "make World" gets all the makefiles built, the dependencies done, and the includes included, and then sits down to do the actual compilations. At this point the machine is running 4 copies of "make" ("make World" runs "make ./X" which runs "make X/lib" which runs ...). What seems to happen is that after a dozen or so files have been compiled, the machine begins to sit around for 10 to 20 seconds between each "rm" and each "cc" in the lowest level makefile. /com/dspst -a shows a process name "--nil--", one of the copies of "make" getting about 10% of the CPU, and about 100 pages/sec of network traffic -- mostly page requests. The page purifiers aren't running though, so it doesn't seem likely that the machine is page thrashing. More likely, its reading stuff out of the X source code directories ... but what? After a while, the machine will eventually crash (which is why I'm currently running the "make" on a diskless workstation!). Has anyone seen this sort of behaviour? Do you know how to get around it? -- David Krowitz krowitz@richter.mit.edu (18.83.0.109) krowitz%richter.mit.edu@eddie.mit.edu krowitz%richter.mit.edu@mitvma.bitnet (in order of decreasing preference)
a20@nikhefh.nikhef.nl (Marten Terpstra) (11/03/90)
In article <9010292020.AA26213@richter.mit.edu> krowitz@RICHTER.MIT.EDU (David Krowitz) writes: >Well, I'm trying to building the X11 R4 sources on our DN3000's, 3500's, and >2500's without much luck. >it doesn't seem likely that the machine is page thrashing. More likely, its >reading stuff out of the X source code directories ... but what? After a while, >the machine will eventually crash (which is why I'm currently running the >"make" on a diskless workstation!). Has anyone seen this sort of behaviour? >Do you know how to get around it? We have seen the same behaviour. The problem was that if you do not have all your files on the machine you're compling X11R4 on, the Apollo will take ages checking the dependencies of the include files. For this it will have to get the timestamp of these files. So it will start swapping and the network activity will go up significantly and the Apollo will eventually die. The solution we used was to get rid off the dependencies. Also you might put all the necessary files on a local disk. This way the Apollo will not use the network and will not disrupt your other Apollo's. Also some Apollos have some difficulties with the optimizer. I compiled it on an Apollo 2500 mith 8Mb memory with all X related files on local disk without the optimizer and without dependency checking and this worked OK. Good luck, Marten -- Marten Terpstra National Institute for Nuclear Internet : terpstra@nikhef.nl and High Energy Physics Oldie-net: {....}mcsun!nikhefh!terpstra (NIKHEF-H), PO Box 41882, 1009 DB Phone : +31 20 592 5102 Amsterdam, The Netherlands