From owner-freebsd-questions Fri Apr 14 05:19:50 1995 Return-Path: questions-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id FAA15620 for questions-outgoing; Fri, 14 Apr 1995 05:19:50 -0700 Received: from crow.csrv.uidaho.edu (root@crow.csrv.uidaho.edu [129.101.119.223]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id FAA15614 for ; Fri, 14 Apr 1995 05:19:48 -0700 Received: from pain.csrv.uidaho.edu (pain.csrv.uidaho.edu [129.101.114.109]) by crow.csrv.uidaho.edu (8.6.9/8.6.9) with ESMTP id FAA26956 for ; Fri, 14 Apr 1995 05:19:40 -0700 Message-Id: <199504141219.FAA26956@crow.csrv.uidaho.edu> To: questions@freefall.cdrom.com Subject: [2.0R] news server memleak? X-Real-Name: Faried Nawaz X-Address: Box 3582, Moscow, Id 83843-1914 X-Phone: 1 (208) 882 8896 Date: Fri, 14 Apr 1995 05:19:38 -0700 From: Faried Nawaz Sender: questions-owner@FreeBSD.org Precedence: bulk hello, i have a freebsd machine that acts as the news server for my campus. i'm running inn1.4 (from the ports) on a micron p90 and 2.0-RELEASE. i've installed a few patches to speed up innxmit, innd, etc...nothing unusual. it has 32mb ram; i reserved 128mb for swap (though i've never seen it above 30 or so); it has a simple vga video card and has two vtys. the scsi controller is a bt946c; the drives are a 500mb quantum and a 2gig barracuda. the machine & os appear to do quite well -- they've impressed my employers. i've seen 20+ day uptimes; its response is quite spiffy, and it even managed to impress the local novell dude (mainly due to the install disks :). i've noticed a strange thing, though, and it puzzles me. yesterday (18 day uptime) i noticed that the machine was living in swap. it had innd running (ps said vsz was around 7.5mb), and had a few in.nnrpds running (ps said vsz for each was just over a meg). i could not see how it all added up to > 20mb, but i was 10mb into swap. last night (a few hours ago, actually), i shutdown innd, kill -15'd the few running nnrpds, and...found myself 3mb into swap. i ran a few commands, and i produce their output below. i'm not quite sure how to interpret the results (or even if i was using the right commands to pinpoint the problem), though. help? ( uptime is 19 days, 3 hours at the time of recording. kern.maxvnodes = 2188 kern.maxproc = 1044 kern.maxfiles = 3404 hw.physmem = 33161216 hw.usermem = 26656768 maxusers (in the kernel config file) is set to 64. i changed CHILD_MAX in /sys/sys/syslimits to 125; news couldn't fork a lot of nnrpds...and this is a large campus (is there a better way?). ) Script started on Fri Apr 14 04:40:34 1995 //newshound~# pstat -T 72/3404 files 2186 vnodes 3M/128M swap space //newshound~# swapinfo -k Device 1K-blocks Used Available Capacity /dev/sd0b 131072 3668 127404 3% //newshound~# vmstat -s 140205270 cpu context switches 517500840 device interrupts 47827105 software interrupts 27898289 traps 201589844 system calls 246510 swap pager pageins 619884 swap pager pages paged in 307442 swap pager pageouts 782911 swap pager pages paged out 127496 vnode pager pageins 487600 vnode pager pages paged in 0 vnode pager pageouts 0 vnode pager pages paged out 43640656 VM object cache lookups 43065520 VM object hits 61694 page daemon wakeups 50238478 pages examined by the page daemon 502501 pages reactivated 28207 intransit blocking page faults 447946629 zero fill pages allocated 9533202 zero fill pages zeroed 4553831 copy-on-write faults 28647694 total VM faults taken 32189702 pages freed 1017555 pages freed by daemon 12739142 pages freed by exiting processes 1289 pages active 187 pages inactive 1657 pages wired down 4498 pages free 4096 bytes per page 50747622 total name lookups cache hits (46% pos + 0% neg) system 27% per-process deletions 0%, falsehits 0%, toolong 0% //newshound~# vmstat -m Memory statistics by bucket size Size In Use Free Requests HighWater Couldfree 16 1162 630 23571935 1280 0 32 1372 3748 13270541 640 0 64 2369 2111 19052481 320 57528 128 3343 1425 173047271 160 0 256 2257 575 25081958 80 0 512 18 6 1673 40 0 1K 8 60 27976272 20 17931306 2K 7 9 126 10 0 4K 8 2 48 5 0 8K 4 5 5561 5 0 16K 0 0 8 5 0 Memory usage type by bucket size Size Type(s) 16 devbuf, routetbl, vnodes, VM objhash, VM pgdata, proc, temp 32 devbuf, pcb, routetbl, pgrp, session, VM mapent, VM pager, VM pgdata, subproc, LFS segment, in_multi, ether_multi, temp 64 devbuf, routetbl, ifaddr, namecache, VM pgdata, file, lockf, LFS segment, temp 128 mbuf, devbuf, pcb, routetbl, fragtbl, zombie, ifaddr, soopts, cred, vnodes, VM map, VM object, VM pgdata, file desc, temp, ttys 256 devbuf, socket, pcb, vnodes, VM map, VM pgdata, file desc, proc, subproc, FFS node, temp 512 mbuf, ioctlops, mount, UFS mount, VM pgdata, file desc, temp 1K devbuf, namei, UFS mount, VM pgdata, temp 2K devbuf, UFS mount, VM pgdata, ttys 4K devbuf, ioctlops, VM pgdata 8K namecache, UFS quota, UFS mount, VM pgdata, temp 16K VM pgdata Memory statistics by type Type Kern Type InUse MemUse HighUse Limit Requests Limit Limit Size(s) mbuf 21 3K 52K 9831K155026281 0 0 128,512 devbuf 15 37K 45K 9831K 63 0 0 16,32,64,128,256,1K,2K,4K socket 41 11K 63K 9831K 326097 0 0 256 pcb 62 11K 47K 9831K 407614 0 0 32,128,256 routetbl 190 15K 16K 9831K 42522 0 0 16,32,64,128 fragtbl 0 0K 1K 9831K 8 0 0 128 zombie 0 0K 1K 9831K 260182 0 0 128 ifaddr 6 1K 1K 9831K 6 0 0 64,128 soopts 2 1K 1K 9831K 7048 0 0 128 namei 0 0K 12K 9831K 27971963 0 0 1K ioctlops 0 0K 4K 9831K 5 0 0 512,4K cred 8 1K 11K 9831K 54477 0 0 128 pgrp 15 1K 1K 9831K 10303 0 0 32 session 12 1K 1K 9831K 8223 0 0 32 mount 5 3K 3K 9831K 5 0 0 512 vnodes 2214 276K 282K 9831K 11091 0 0 16,128,256 namecache 2189 145K 145K 9831K 2189 0 0 64,8K UFS quota 1 8K 8K 9831K 1 0 0 8K UFS mount 13 23K 23K 9831K 13 0 0 512,1K,2K,8K VM map 24 6K 25K 9831K 1449934 0 0 128,256 VM mapent 325 11K 52K 9831K 262009 0 0 32 VM object 833 105K 223K 9831K 15784314 0 0 128 VM objhash 569 9K 11K 9831K 11781486 0 0 16 VM pager 704 22K 45K 9831K 12093352 0 0 32 VM pgdata 839 35K 178K 9831K 12405218 0 0 16,32,64,128,256,512,1K,2K,4K,8K,16K file 72 5K 27K 9831K 15671661 0 0 64 file desc 25 5K 20K 9831K 262991 0 0 128,256,512 lockf 1 1K 1K 9831K 2692926 0 0 64 proc 23 6K 24K 9831K 260264 0 0 16,256 subproc 19 1K 3K 9831K 260829 0 0 32,256 LFS segment 0 0K 4K 9831K 795729 0 0 32,64 FFS node 2117 530K 547K 9831K 24148614 0 0 256 in_multi 2 1K 1K 9831K 2 0 0 32 ether_multi 1 1K 1K 9831K 1 0 0 32 temp 99 11K 27K 9831K 8896 0 0 16,32,64,128,256,512,1K,8K ttys 101 15K 16K 9831K 1561 0 0 128,2K Memory Totals: In Use Free Requests 1287K 710K 282007878 //newshound~# top -b load averages: 0.00, 0.02, 0.00 04:40:47 17 processes: 1 running, 16 sleeping Memory: Real: 1048K/13276K Virt: 36160K/317024K Free: 17992K PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND 21045 root 18 0 872K 944K sleep 0:00 0.20% 0.10% bash 66 root 2 0 192K 184K sleep 3:58 0.05% 0.05% routed 21050 root 30 0 292K 684K run 0:00 0.00% 0.00% top 20943 root 2 0 204K 512K sleep 0:00 0.00% 0.00% rlogind 20955 root 18 0 876K 920K sleep 0:00 0.00% 0.00% bash 20944 fn 18 0 440K 264K sleep 0:00 0.00% 0.00% csh 59 root 18 0 376K 220K sleep 0:58 0.00% 0.00% cron 1 root 10 0 404K 72K sleep 0:02 0.00% 0.00% init 21043 root 3 0 220K 472K sleep 0:00 0.00% 0.00% script 21044 root 3 0 220K 236K sleep 0:00 0.00% 0.00% script 29861 root 3 0 152K 228K sleep 0:00 0.00% 0.00% getty 21641 root 3 0 152K 228K sleep 0:00 0.00% 0.00% getty 69 root 2 0 840K 784K sleep 3:02 0.00% 0.00% named 48 root 2 0 192K 304K sleep 12:53 0.00% 0.00% syslogd 62 daemon 2 0 204K 272K sleep 0:01 0.00% 0.00% portmap //newshound~# vmstat procs memory page disks faults cpu r b w avm fre flt re pi po fr sr f0 s0 s1 in sy cs us sy id 1 0 0 36052 17976 17 0 0 0 19 30 0 4 29 313 122 85 3 3 94 //newshound~# ps alxww UID PID PPID CPU PRI NI VSZ RSS WCHAN STAT TT TIME COMMAND 0 0 0 0 -18 0 0 0 sched DLs ?? 0:00.00 (swapper) 0 1 0 0 10 0 404 72 wait Is ?? 0:02.30 /sbin/init -- 0 2 0 19 -18 0 0 12 psleep DL ?? 10:08.63 (pagedaemon) 0 3 0 0 -6 0 0 12 update DL ?? 16:24.08 (update) 0 48 1 0 2 0 192 304 select Ss ?? 12:53.59 syslogd 0 59 1 0 18 0 376 220 pause Is ?? 0:58.16 cron 1 62 1 0 2 0 204 272 select Is ?? 0:01.37 portmap 0 66 1 0 2 0 192 184 select Ss ?? 3:58.80 routed -q 0 69 1 0 2 0 840 784 select Is ?? 3:02.48 named 0 82 1 0 2 0 424 232 netcon Is ?? 0:06.81 sendmail: accepting connections (sendmail) 0 85 1 0 2 0 244 268 select Is ?? 0:02.15 inetd 0 20943 85 0 2 0 204 512 select S ?? 0:00.74 rlogind 1001 20944 20943 0 18 0 440 264 pause Is p0 0:00.17 -csh (csh) 0 20955 20944 0 18 0 876 920 pause S p0 0:00.93 -su (bash) 0 21043 20955 0 3 0 220 472 ttyin S+ p0 0:00.05 script 0 21044 21043 0 3 0 220 236 ttyin S+ p0 0:00.03 script 0 21045 21044 1 18 0 872 944 pause Ss p1 0:00.16 sh -i (bash) 0 21052 21045 2 28 0 436 240 - R+ p1 0:00.00 ps alxww 0 21641 1 6 3 0 152 228 ttyin Is+ v0 0:00.41 /usr/libexec/getty Pc ttyv0 0 29861 1 2 3 0 152 228 ttyin Is+ v1 0:00.08 /usr/libexec/getty Pc ttyv1 //newshound~# exit exit Script done on Fri Apr 14 04:41:02 1995 thanks for your time, faried. -- fn@uidaho.edu fn::=`faried nawaz`, uidaho::=`university of idaho, moscow, id 83844`, disclaimer::=`my parents, bosses, and friends may or may not agree with this message`.