Adrian Chadd wrote:
> Kinkie, what are you afraid of? The stuff that Henrik and I are
> working on wont' even appear in 2.5. 2.5 was slated to be another
> "feature++" release. I wouldn't mind it being called 2.4.1 or something,
> but quite a lot of low-level stuff (eg cbdata) has changed between
> 2.4->2.5 to justify the version bump.
More importantly: squid.conf format has changed.
> So, don't worry about the 2.5 branch code becoming unstable.
> I'm personally not looking to break squid-HEAD until 2.5 heads out
> the door.
The code I am working on now is 3.0 stuff I think..
> On that track, what are the final issues with 2.4 ? I can't put my
> finger on the poll issue, since I'm just not a linux person.
> Anything else?
There are several seg-faults reported on squid-users.. and a build
problem on IRIX it seems..
/Henrik
Return-Path: <ales@megared.net.mx>
Received: from localhost (IDENT:root@localhost [127.0.0.1])
by henrik.localdomain (8.9.3/8.9.3) with ESMTP id VAA31581
for <henrik@localhost>; Tue, 27 Feb 2001 21:46:55 +0100
Received: from pop.passagen.se
by localhost with POP3 (fetchmail-5.3.1)
for henrik@localhost (single-drop); Tue, 27 Feb 2001 21:46:55 +0100 (CET)
Received: from unix.megared.net.mx (megamail.megared.com.mx [200.52.207.52]) by mail1.passagen.se (8.9.3/8.9.3/1.22)
with ESMTP id <UAA09914> for <hno@hem.passagen.se>; Tue, 27 Feb 2001 20:34:24 +0100 (MET)
Received: from Alesito (ales.corp.megared.net.mx [200.52.193.2])
by unix.megared.net.mx (8.11.0/8.11.0) with SMTP id f1RJUHD52744;
Tue, 27 Feb 2001 13:30:17 -0600 (CST)
(envelope-from ales@megared.net.mx)
Message-ID: <07eb01c0a0f4$3e3503a0$d3620a0a@megared.net.mx>
From: "Alejandro A. Ramirez" <ales@megared.net.mx>
To: "Henrik Nordstrom" <hno@hem.passagen.se>
Cc: <squid-users@ircache.net>
Subject: Squid 2.4 assertion failed: store_dir_diskd.c:1630
Date: Tue, 27 Feb 2001 13:34:03 -0600
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.50.4133.2400
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
Hi Henrik,
Using diskd again (and the 2 patches that you sent to me, also
"pipeline_prefetch off" in squid.conf) in the server I hitted another error:
> 2001/02/21 19:00:55| WARNING: failed to unpack meta data
> 2001/02/21 19:01:02| WARNING: failed to unpack meta data
> squid in free(): warning: modified (chunk-) pointer.
> 2001/02/21 19:01:08| assertion failed: store_dir_diskd.c:1630: "0 <=
> *shm_offset && *shm_offset < SHMBUFS
> * SHMBUF_BLKSZ"
>
> Program received signal SIGABRT, Aborted.
> 0x4816a3bc in kill () from /usr/lib/libc.so.4
> (gdb) bt
> #0 0x4816a3bc in kill () from /usr/lib/libc.so.4
> #1 0x481a6026 in abort () from /usr/lib/libc.so.4
> #2 0x806753b in xassert (
> msg=0x80e2940 "0 <= *shm_offset && *shm_offset < SHMBUFS *
> SHMBUF_BLKSZ",
> file=0x80e1764 "store_dir_diskd.c", line=1630) at debug.c:250
> #3 0x80b7a07 in storeDiskdShmGet (sd=0x820d948, shm_offset=0xbfbbf908)
> at store_dir_diskd.c:1630
> #4 0x80b8ccd in storeDiskdRead (SD=0x820d948, sio=0x1989e600,
> buf=0x244a0000 "", size=4096, offset=24576,
> callback=0x80a4de8 <storeClientReadBody>, callback_data=0x224fc780)
> at store_io_diskd.c:214
> #5 0x80a4185 in storeRead (sio=0x1989e600, buf=0x244a0000 "", size=4096,
> offset=24576, callback=0x80a4de8 <storeClientReadBody>,
> callback_data=0x224fc780) at store_io.c:82
> #6 0x80a4ddb in storeClientFileRead (sc=0x224fc780) at store_client.c:348
> #7 0x80a4c68 in storeClientCopy3 (e=0x1aaaa080, sc=0x224fc780)
> at store_client.c:328
> #8 0x80a4972 in storeClientCopy2 (e=0x1aaaa080, sc=0x224fc780)
> at store_client.c:261
> #9 0x80a47aa in storeClientCopy (sc=0x224fc780, e=0x1aaaa080,
> seen_offset=24455, copy_offset=24455, size=4096, buf=0x244a0000 "",
> callback=0x805f400 <clientSendMoreData>, data=0x1dd97400)
> at store_client.c:210
> ---Type <return> to continue, or q <return> to quit---
> #10 0x80600f3 in clientWriteComplete (fd=4971, bufnotused=0x0, size=4096,
> errflag=0, data=0x1dd97400) at client_side.c:1938
> #11 0x805fb74 in clientWriteBodyComplete (fd=4971,
> buf=0x24b48000 "\xa9\024q\"\225\a}\xf3\226\xf9\xb6OGPh
> 5\xb9\n\214\xeb\xd3\xe0 \xef\xcb:}\xbc\204\xdb
>
\xf6\xf1\xfc3\223y\xa0\xc96\xb0\xe6*\031\xfd8\xf9\203\xb2\xf0\xa7o|\xa4l\234
> \xdc\xaf\xde\227\xa3B\xf1F\xc
> 4\xd2H\xc9i", size=4096, errflag=0,
> data=0x1dd97400) at client_side.c:1833
> #12 0x8063464 in CommWriteStateCallbackAndFree (fd=4971, code=0) at
> comm.c:98
> #13 0x8065582 in commHandleWrite (fd=4971, data=0x108c89c0) at comm.c:858
> #14 0x8066564 in comm_poll (msec=0) at comm_select.c:446
> #15 0x808836d in main (argc=2, argv=0xbfbffc24) at main.c:720
> #16 0x804a529 in _start ()
Greetings
Ales
Return-Path: <taver@otenet.gr>
Received: from localhost (IDENT:root@localhost [127.0.0.1])
by henrik.localdomain (8.9.3/8.9.3) with ESMTP id WAA27656
for <henrik@localhost>; Mon, 26 Feb 2001 22:50:48 +0100
Received: from pop.passagen.se
by localhost with POP3 (fetchmail-5.3.1)
for henrik@localhost (single-drop); Mon, 26 Feb 2001 22:50:48 +0100 (CET)
Received: from noc.otenet.gr (noc.otenet.gr [195.170.0.29]) by mail1.passagen.se (8.9.3/8.9.3/1.22)
with ESMTP id <VAA17544> for <hno@hem.passagen.se>; Mon, 26 Feb 2001 21:36:26 +0100 (MET)
Received: (from taver@localhost)
by noc.otenet.gr (8.10.2/8.11.1) id f1QKaOd23307
for hno@hem.passagen.se; Mon, 26 Feb 2001 22:36:24 +0200 (EET)
Date: Mon, 26 Feb 2001 22:36:24 +0200
From: Costas Tavernarakis <taver@otenet.gr>
To: Henrik Nordstrom <hno@hem.passagen.se>
Subject: Squid 2.4 bugs
Message-ID: <20010226223624.A13724@noc.otenet.gr>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 1.0.1i
Precedence: first-class
Hi,
I just had a look for the solaris-related bugs in squid 2.4 at the
bugzilla on squid-cache.org:
a. Squid "NULL pointer access in storeDirSwapLog while writing icons" bug
(bug-ID: 79).
This is caused by the :25 and :7 size definitions in struct _StoreEntry
(lines 1323,1324 in structs.h). It seems all Sun C compilers are not
happy with these (at least cc 5.0 and 6.1 on sunos 5.8).
I see (from cvsweb on squid-cache.org) that it was your change...
I tried to revert this by changing them both to be "int", with success.
Maybe more memory-efficient definitions can be made (e.g a "char" definition
for swap_dirno), but i don't know about allignment issues.
So this bug can be closed with a bit of work.
I WAS involved with noting/reporting this bug in the past, and i know
i should have emailed Duane about this again, but it looks like he
ignores my messages about tracing this issue...
b. About "diskd not working on Solaris" bugs (bug #13, #86):
As M.A.Young@durham.ac.uk points out, the fcntl argument on line
314 in diskd.c should be changed from SQUID_NONBLOCK to O_NONBLOCK.
This change is suffient and works. So, this bug can be closed, too.
c. Bug "#93" is autoconf stuff
I hope i've helped bringing squid-2.4 a little closer to release...
-- ======================================================================== Costas Tavernarakis OTEnet S.A. Internet Service Provider E-mail: taver@otenet.gr Work Phone: 01-3826199 ========================================================================
Return-Path: <squid-users-request@ircache.net>
Received: from localhost (IDENT:root@localhost [127.0.0.1])
by henrik.localdomain (8.9.3/8.9.3) with ESMTP id IAA22829
for <henrik@localhost>; Wed, 14 Feb 2001 08:34:42 +0100
Received: from pop.passagen.se
by localhost with POP3 (fetchmail-5.3.1)
for henrik@localhost (single-drop); Wed, 14 Feb 2001 08:34:42 +0100 (CET)
Received: from jurongcement.com (ad202.166.79.7.magix.com.sg [202.166.79.7]) by mail1.passagen.se (8.9.3/8.9.3/1.22)
with ESMTP id <XAA03907> for <hno@hem.passagen.se>; Tue, 13 Feb 2001 23:28:54 +0100 (MET)
Received: from mail.jurongcement.com (account jurongcement)
by jurongcement.com (CommuniGate Pro RPOP 3.2.4)
with RPOP id 971152; Wed, 14 Feb 2001 06:38:15 +0800
Received: from ircache.net (ircache.net [206.168.0.6])
by www.d-ns.net (8.9.3/8.9.3) with ESMTP id EAA07741
for <arthurtan@jurongcement.com>; Wed, 14 Feb 2001 04:40:31 +0800
Received: (from slist@localhost)
by ircache.net (8.11.1/8.9.3) id f1DKbcl70744;
Tue, 13 Feb 2001 13:37:38 -0700 (MST)
(envelope-from squid-users-request@ircache.net)
Resent-Date: Tue, 13 Feb 2001 13:26:31 -0700 (MST)
X-Envelope-From: ales@megared.net.mx Tue Feb 13 13:26:31 2001
Received: from unix.megared.net.mx (megamail.megared.com.mx [200.52.207.52])
by ircache.net (8.11.1/8.9.3) with ESMTP id f1DKQUN67706
for <squid-users@ircache.net>; Tue, 13 Feb 2001 13:26:30 -0700 (MST)
(envelope-from ales@megared.net.mx)
Received: from Alesito (ales.corp.megared.net.mx [200.52.193.2])
by unix.megared.net.mx (8.11.0/8.11.0) with SMTP id f1DKPYN92722;
Tue, 13 Feb 2001 14:25:35 -0600 (CST)
(envelope-from ales@megared.net.mx)
Message-ID: <01e801c095fb$2001e960$d3620a0a@megared.net.mx>
From: "Alejandro A. Ramirez" <ales@megared.net.mx>
To: "Henrik Nordstrom" <hno@hem.passagen.se>
Cc: <squid-users@ircache.net>
Date: Tue, 13 Feb 2001 14:25:36 -0600
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.50.4133.2400
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
Subject: [SQU] Squid 2.4 Head FATAL: Received Segment Violation...dying
Resent-Message-ID: <GOieRC.A.JiQ.3hZi6@ircache.net>
Resent-From: squid-users@ircache.net
X-Mailing-List: <squid-users@ircache.net> archive/latest/1317
X-Loop: squid-users@ircache.net
Precedence: list
Resent-Sender: squid-users-request@ircache.net
Hi Henrik,
Im testing the squid-2.4-200102120000 version, and it seems that I have
hitted a bug, below you will find the backtrace, Im using FreeBSD 4.2 with 2
PIII 600 Mhz processors, and 1 Gig of Ram, it was serving 18.7 kreq/min with
a very good performance, the configuration options are as follows:
setenv PREFIX /usr/local
./configure --prefix=${PREFIX} \
--with-layout=GNU \
--bindir=${PREFIX}/sbin \
--sysconfdir=${PREFIX}/etc/squid \
--localstatedir=${PREFIX}/squid \
--enable-snmp \
--disable-ident-lookups \
--enable-underscores \
--disable-icp \
--enable-truncate \
--enable-storeio=ufs,diskd
And in the squid.conf I have all the cache_dirīs like this one :
cache_dir diskd -1 /cache0/0 6000 16 256 64 72
The kernel is compiled with the following options:
options NMBCLUSTERS=32768 #Support for 32M for MBufs
options MAXDSIZ="(1024*1024*1024)"
options DFLDSIZ="(1024*1024*1024)"
options PQ_LARGECACHE
options SYSVMSG
options MAXFILES=32768
options MSGMNB=16384
options MSGMNI=41
options MSGSEG=2049
options MSGSSZ=64
options MSGTQL=1024
options SYSVSHM
options SHMSEG=16
options SHMMNI=32
options SHMMAX=2097152
options SHMALL=8192
The system log file it showed this:
Feb 13 12:35:22 cache /kernel: pid 214 (squid), uid 80: exited on signal 11
cache# netstat -mb
68/16352/131072 mbufs in use (current/peak/max):
66 mbufs allocated to data
2 mbufs allocated to packet headers
64/5632/32768 mbuf clusters in use (current/peak/max)
15352 Kbytes allocated to network (15% of mb_map in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines
cache# df
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/idad0s1a 254063 43537 190201 19% /
/dev/idad0s1h 60521954 619863 55060335 1% /cache0
/dev/da0s1e 17235871 577908 15279094 4% /cache1/0
/dev/da1s1e 17235871 567513 15289489 4% /cache1/1
/dev/da2s1e 17235871 629858 15227144 4% /cache1/2
/dev/da3s1e 17235871 619487 15237515 4% /cache1/3
/dev/da4s1e 17235871 586776 15270226 4% /cache1/4
/dev/da5s1e 17235871 577241 15279761 4% /cache1/5
/dev/idad0s1e 1016303 11 934988 0% /tmp
/dev/idad0s1g 4065262 544458 3195584 15% /usr
/dev/idad0s1f 2032623 599 1869415 0% /var
procfs 4 4 0 100% /proc
The filesystems are mounted with softupdates & noatime option.
Here are the first two errors that I got before I used GDB for getting a
backtrace:
2001/02/13 11:26:33| clientReadRequest: FD 672 Invalid Request
2001/02/13 11:26:33| clientSendMoreData: Deferring error:invalid-request
FATAL: Received Segment Violation...dying.
2001/02/13 11:26:38| storeDirWriteCleanLogs: Starting...
2001/02/13 11:26:38| WARNING: Closing open FD 22
2001/02/13 11:26:38| 65536 entries written so far.
2001/02/13 11:26:38| 131072 entries written so far.
2001/02/13 11:26:38| Finished. Wrote 156863 entries.
2001/02/13 11:26:38| Took 0.5 seconds (325129.6 entries/sec).
CPU Usage: 2526.218 seconds = 1370.164 user + 1156.054 sys
Maximum Resident Size: 245556 KB
Page faults with physical i/o: 0
2001/02/13 11:30:34| clientReadRequest: FD 4339 Invalid Request
2001/02/13 11:30:34| clientSendMoreData: Deferring error:invalid-request
FATAL: Received Bus Error...dying.
2001/02/13 11:30:36| storeDirWriteCleanLogs: Starting...
2001/02/13 11:30:36| WARNING: Closing open FD 28
2001/02/13 11:30:36| 65536 entries written so far.
2001/02/13 11:30:36| 131072 entries written so far.
2001/02/13 11:30:36| Finished. Wrote 168271 entries.
2001/02/13 11:30:36| Took 0.4 seconds (384771.6 entries/sec).
CPU Usage: 183.611 seconds = 97.945 user + 85.666 sys
Maximum Resident Size: 154836 KB
Page faults with physical i/o: 2
Here it is the GDB Backtrace:
Script started on Tue Feb 13 12:01:17 2001
GNU gdb 4.18
Copyright 1998 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show warranty" for details.
This GDB was configured as "i386-unknown-freebsd"...
(gdb) handle SIGPIPE noprint nostop passhandle SIGPIPE noprint nostop pass
Signal Stop Print Pass to program Description
SIGPIPE No No Yes Broken pipe
(gdb) r -CNd1r -CNd1
Starting program: /usr/local/sbin/squid -CNd1
2001/02/13 12:01:28| Starting Squid Cache version 2.4.PRE-STABLE for
i386-unknown-freebsd4.2...
2001/02/13 12:01:28| Process ID 214
2001/02/13 12:01:28| With 32768 file descriptors available
2001/02/13 12:01:28| Performing DNS Tests...
2001/02/13 12:01:28| Successful DNS name lookup tests...
2001/02/13 12:01:28| DNS Socket created on FD 7
2001/02/13 12:01:28| Adding nameserver 127.0.0.1 from /etc/resolv.conf
2001/02/13 12:01:28| Unlinkd pipe opened on FD 12
2001/02/13 12:01:28| Swap maxSize 43008000 KB, estimated 3308307 objects
2001/02/13 12:01:28| Target number of buckets: 165415
2001/02/13 12:01:28| Using 262144 Store buckets
2001/02/13 12:01:28| Max Mem size: 131072 KB
2001/02/13 12:01:28| Max Swap size: 43008000 KB
2001/02/13 12:01:28| Store logging disabled
2001/02/13 12:01:28| Rebuilding storage in /cache0/0 (DIRTY)
2001/02/13 12:01:28| Rebuilding storage in /cache1/0/0 (DIRTY)
2001/02/13 12:01:28| Rebuilding storage in /cache1/1/0 (DIRTY)
2001/02/13 12:01:28| Rebuilding storage in /cache1/2/0 (DIRTY)
2001/02/13 12:01:28| Rebuilding storage in /cache1/3/0 (DIRTY)
2001/02/13 12:01:28| Rebuilding storage in /cache1/4/0 (DIRTY)
2001/02/13 12:01:28| Rebuilding storage in /cache1/5/0 (DIRTY)
2001/02/13 12:01:28| Using Round Robin store dir selection
2001/02/13 12:01:28| Set Current Directory to /cache0/0
2001/02/13 12:01:28| Loaded Icons.
2001/02/13 12:01:28| Accepting HTTP connections at 0.0.0.0, port 80, FD 31.
2001/02/13 12:01:28| Accepting SNMP messages on port 1610, FD 33.
2001/02/13 12:01:28| Accepting WCCP messages on port 2048, FD 34.
2001/02/13 12:01:28| Ready to serve requests.
2001/02/13 12:01:30| Store rebuilding is 17.0% complete
2001/02/13 12:01:37| Done reading /cache1/1/0 swaplog (24037 entries)
2001/02/13 12:01:37| Done reading /cache1/0/0 swaplog (24078 entries)
2001/02/13 12:01:37| Done reading /cache1/2/0 swaplog (24075 entries)
2001/02/13 12:01:37| Done reading /cache1/3/0 swaplog (24077 entries)
2001/02/13 12:01:37| Done reading /cache1/5/0 swaplog (24077 entries)
2001/02/13 12:01:37| Done reading /cache0/0 swaplog (24114 entries)
2001/02/13 12:01:37| Done reading /cache1/4/0 swaplog (24142 entries)
2001/02/13 12:01:37| Finished rebuilding storage from disk.
2001/02/13 12:01:37| 168360 Entries scanned
2001/02/13 12:01:37| 0 Invalid entries.
2001/02/13 12:01:37| 0 With invalid flags.
2001/02/13 12:01:37| 168117 Objects loaded.
2001/02/13 12:01:37| 0 Objects expired.
2001/02/13 12:01:37| 240 Objects cancelled.
2001/02/13 12:01:37| 4 Duplicate URLs purged.
2001/02/13 12:01:37| 2 Swapfile clashes avoided.
2001/02/13 12:01:37| Took 8.4 seconds (19971.4 objects/sec).
2001/02/13 12:01:37| Beginning Validation Procedure
2001/02/13 12:01:37| Completed Validation Procedure
2001/02/13 12:01:37| Validated 168114 Entries
2001/02/13 12:01:37| store_swap_size = 1619718k
2001/02/13 12:01:37| storeLateRelease: released 0 objects
2001/02/13 12:01:48| parseHttpRequest: Unsupported method 'HA'
2001/02/13 12:01:48| clientReadRequest: FD 150 Invalid Request
2001/02/13 12:01:58| parseHttpRequest: Requestheader contains NULL
characters
...
...
...
2001/02/13 12:27:54| parseHttpRequest: Requestheader contains NULL
characters
2001/02/13 12:27:54| clientReadRequest: FD 794 Invalid Request
2001/02/13 12:27:54| clientSendMoreData: Deferring error:invalid-request
Program received signal SIGSEGV, Segmentation fault.
0x80aa847 in linklistPush (L=0x132843, p=0x50137000) at tools.c:873
873 while (*L)
(gdb) bt
#0 0x80aa847 in linklistPush (L=0x132843, p=0x50137000) at tools.c:873
#1 0x80b7a20 in storeDiskdShmPut (sd=0x820d948, offset=192512) at
store_dir_diskd.c:1643
#2 0x80b482c in storeDiskdDirCallback (SD=0x820d948) at
store_dir_diskd.c:532
#3 0x80a6ac3 in storeDirCallback () at store_dir.c:463
#4 0x8065f0a in comm_poll (msec=0) at comm_select.c:330
#5 0x80882ed in main (argc=2, argv=0xbfbffc24) at main.c:720
#6 0x804a529 in _start ()
(gdb) c
Continuing.
Program terminated with signal SIGSEGV, Segmentation fault.
The program no longer exists.
Script done on Tue Feb 13 12:35:41 2001
I Hope this can help you to debug this, if there is anything else that I can
do to help, just let me know.
Greetings
Ales
-- To unsubscribe, see http://www.squid-cache.org/mailing-lists.htmlReceived on Tue Feb 27 2001 - 14:15:17 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:13:35 MST