Date: Fri, 23 Jun 2000 17:19:43 -0700 (PDT) From: wessels@ircache.net To: freebsd-gnats-submit@FreeBSD.org Subject: kern/19479: processes stuck in 'ffsvgt' and 'FFS no' states Message-ID: <20000624001943.570B937BAB1@hub.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 19479
>Category: kern
>Synopsis: processes stuck in 'ffsvgt' and 'FFS no' states
>Confidential: no
>Severity: serious
>Priority: low
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Fri Jun 23 17:20:01 PDT 2000
>Closed-Date:
>Last-Modified:
>Originator: Duane Wessels
>Release: RELENG_3 from 20000620
>Organization:
IRCache
>Environment:
FreeBSD mr-garrison.measurement-factory.com 3.5-STABLE FreeBSD 3.5-STABLE #4: Tue Jun 20 14:15:04 MDT 2000
root@mr-garrison.measurement-factory.com:/usr/src/sys/compile/SQUID i386
>Description:
I'm benchmarking the Squid proxy cache, which generates a lot
of disk I/O. After an hour or two, all of the Squid processes
get stuck in the 'ffsvgt' state, and one of them is stuck
in 'FFS no':
PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND
368 wessels 98 0 102M 99036K RUN 106:01 98.14% 98.14% squid
375 wessels -18 0 1272K 812K ffsvgt 2:23 0.00% 0.00% diskd
373 wessels -18 0 1272K 812K ffsvgt 2:22 0.00% 0.00% diskd
374 wessels -20 0 1272K 812K FFS no 2:22 0.00% 0.00% diskd
372 wessels -18 0 1272K 812K ffsvgt 2:20 0.00% 0.00% diskd
370 wessels -18 0 1272K 812K ffsvgt 2:19 0.00% 0.00% diskd
371 wessels -18 0 1272K 812K ffsvgt 2:19 0.00% 0.00% diskd
Although the system continues to run, it seems like any process that
tries to access a filesystem also becomes stuck in the ffsvgt
state. I can't start new commands, log in, etc.
Filesystems are mounted normally. No special options:
/dev/da0c /cache1 ufs rw 0 3
/dev/da1c /cache2 ufs rw 0 3
/dev/da2c /cache3 ufs rw 0 3
/dev/da3c /cache4 ufs rw 0 3
/dev/da4c /cache5 ufs rw 0 3
/dev/da5c /cache6 ufs rw 0 3
I gather that 'FFS no' is related to filesystem vnodes. I wonder
if I'm running out of vnodes, or if they're not getting freed?
Here's systat after the benchmark has been running for 20 minutes:
2 users Load 1.23 1.16 0.90 Fri Jun 23 18:10
Mem:KB REAL VIRTUAL VN PAGER SWAP PAGER
Tot Share Tot Share Free in out in out
Act 68748 1064 74260 3860 21700 count
All 513584 1440 3026816 4408 pages
53 cow Interrupts
Proc:r p d s w Csw Trp Sys Int Sof Flt 75 zfod 2377 total
1 7 4 527 297 6247 2377 268 152 50408 wire 100 clk0 irq0
71716 act 128 rtc0 irq8
45.3%Sys 6.3%Intr 47.1%User 0.0%Nice 1.2%Idl 370040 inact 1864 pci irq10
| | | | | | | | | | 20564 cache 266 pci irq11
=======================+++>>>>>>>>>>>>>>>>>>>>>> 1136 free fdc0 irq6
daefr 19 wdc0 irq14
Namei Name-cache Dir-cache 106 prcfr
Calls hits % hits % react
1920 1529 80 6 0 pdwake
525 pdpgs
Discs da0 da1 da2 da3 da4 da5 wd0 intrn
KB/t 7.18 7.38 8.13 8.26 10.58 8.03 8.00 8342 buf
tps 55 57 50 40 18 47 1 202 dirtybuf
MB/s 0.39 0.41 0.39 0.32 0.18 0.37 0.01 32702 desiredvnodes
% busy 25 24 25 16 12 21 0 55047 numvnodes
24 freevnodes
>How-To-Repeat:
Install the latest squid-2.4 code and pummel it with a
Web Polygraph workload.
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted:
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000624001943.570B937BAB1>