(Illustration by Gaich Muramatsu)
On Tue, Dec 21, 1999 at 03:50:54PM +0100, Detlef Weitz wrote: > 2. portmaplog remains empty, > maybe here the reason of failures? > But rpc2portmapper is running (seems so) No that cannot be the reason. The rpc2portmapper is only used by the updateclnt and updatesrv daemons. > Assertion failed: ret == RVM_SUCCESS, file "recovc.cc", line 168 > Sleeping forever. You may use gdb to attach to process 4914./vice/srv# Earlier you had sent an email about using rdsinit. The parameters you showed seem very low (about 1 or 2MB rvm??). The parameters we have for 22MB RVM for our smallest servers are: You 22MB RVM length of data segment (total rvm size) 0xf5000 0x1600000 starting address of RVM (OS dependent) 0x20000000 0x20000000 heap length (main data allocation part) 0xf0000 0x1200000 static length (used for some globals) 0x4000 0x100000 nlists (number of allocation buckets?) 64 80 chunk size (minimal allocation unit?) 32 32 It could very well be that the server isn't able to run correctly because it cannot fit it's static of heap allocated data in these segments. Basically staticlength + heaplength ~= total rvm size, a little bit less due to some headers and stuff. Maybe if you scale our 22MB rvm parameters to rvmsize 0x800000, heaplen 0x600000, staticlen 0x100000 it might work. No need to mess with the starting address, nlists or chunk size. That's just to tweak the allocator. JanReceived on 1999-12-21 12:24:57