Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
On Fri, Nov 10, 2006, Adrian Chadd wrote: As I said to someone else a few weeks ago - we could fix diskd but it'd require quite a bit of effort and there's other things higher on the priority list for us to fix. .. and I've got a patch to fix this behaviour, in case people are running diskd and see the crash shown in Bug #761. http://www.squid-cache.org/bugs/show_bug.cgi?id=761 Adrian
Re: [squid-users] DISKD Autoremoved
Hi, At 02.53 14/10/2007, Juan C. Crespo R. wrote: hi Does anyone knows why when I try to install squid with diskd, this option autoremove itself? ./configure --prefix=/usr/local/squid --enable-async-io=128 --enable-storio=diskd,ufs --enable- There is a typo: it should be --enable-storeio. Regards Guido - Guido Serassio Acme Consulting S.r.l. - Microsoft Certified Partner Via Lucia Savarino, 1 10098 - Rivoli (TO) - ITALY Tel. : +39.011.9530135 Fax. : +39.011.9781115 Email: [EMAIL PROTECTED] WWW: http://www.acmeconsulting.it/
Re: [squid-users] DISKD Autoremoved
Thank you Guido Serassio escribió: Hi, At 02.53 14/10/2007, Juan C. Crespo R. wrote: hi Does anyone knows why when I try to install squid with diskd, this option autoremove itself? ./configure --prefix=/usr/local/squid --enable-async-io=128 --enable-storio=diskd,ufs --enable- There is a typo: it should be --enable-storeio. Regards Guido - Guido Serassio Acme Consulting S.r.l. - Microsoft Certified Partner Via Lucia Savarino, 1 10098 - Rivoli (TO) - ITALY Tel. : +39.011.9530135 Fax. : +39.011.9781115 Email: [EMAIL PROTECTED] WWW: http://www.acmeconsulting.it/
[squid-users] DISKD Autoremoved
hi Does anyone knows why when I try to install squid with diskd, this option autoremove itself? ./configure --prefix=/usr/local/squid --enable-async-io=128 --enable-storio=diskd,ufs --enable-removal-policies=lru,heap --enable-pf-transparent --with-large-files --enable-delay-pools --with-dl | more checking for a BSD-compatible install... /usr/bin/install -c checking whether build environment is sane... yes checking for gawk... no checking for mawk... no checking for nawk... nawk checking whether make sets $(MAKE)... yes checking whether to enable maintainer-specific portions of Makefiles... no checking for gcc... gcc checking for C compiler default output file name... a.out checking whether the C compiler works... yes checking whether we are cross compiling... no checking for suffix of executables... checking for suffix of object files... o checking whether we are using the GNU C compiler... yes checking whether gcc accepts -g... yes checking for gcc option to accept ISO C89... none needed checking for style of include used by make... GNU checking dependency style of gcc... gcc3 checking whether gcc and cc understand -c and -o together... yes checking build system type... i386-unknown-freebsd6.2 checking host system type... i386-unknown-freebsd6.2 checking for pkg-config... /usr/local/bin/pkg-config With 128 aufs threads With pthreads With dl Store modules built: ufs aufs Removal policies built: lru heap Delay pools enabled PF Transparent Proxy enabled Thanks
[squid-users] diskd question
Greetings, So I am using local disk for my cache. This consists of a 500G SATA drive. My cache size is 50G. I tried using a queue size of Q1=72 and Q1=64, however it looks like I am still I/O constrained with http requests taking up to 11 seconds. I am using UFS. Logging, and access time have been disabled. I am now running at: Q1=12 Q2=10 Does anyone happen to have any suggestions? Thanks!
Re: [squid-users] diskd question
Hi Frank, Frank Ruiz wrote: Greetings, So I am using local disk for my cache. This consists of a 500G SATA drive. My cache size is 50G. I tried using a queue size of Q1=72 and Q1=64, however it looks like I am still I/O constrained with http requests taking up to 11 seconds. Which operating system and squid version are you using diskd with? With my limited experience, diskd is better suited for FreeBSD systems. However, it's performance is more than satisfactory in Linux systems too. Are you using SquidClient or SNMP to measure the http requests response time? Bandwidth saturation and alot of ACL filtering could also slow down your http response time. Posting your squid.conf may help. Also how many users do you have and what is the hardware configuration of your Squid box? What does the following squidclient output say: squidclient mgr:5min | grep client I am using UFS. Logging, and access time have been disabled. I am now running at: Q1=12 Q2=10 I am a little confused. So now you are using Q1=12 Q2=10 for diskd? If you are, I think that they are too small a value. If your Squid box is a busy machine, then disabling logging to access.log or cache.log may help to some extent. Does anyone happen to have any suggestions? You can try something like: cache_dir diskd /cache 5 48 256 Q1=64 Q2=72 By the way, running a 50 GB cache may need alot of memory and a fast hard drive too! Hope it helps. Thanking you... Thanks! -- With best regards and good wishes, Yours sincerely, Tek Bahadur Limbu (TAG/TDG Group) Jwl Systems Department Worldlink Communications Pvt. Ltd. Jawalakhel, Nepal http://www.wlink.com.np
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
On 08.11.06 22:48, Tek Bahadur Limbu wrote: I am using squid-2.5/2.6 with FreeBSD 6.0/6.1 on some of my Dell SC420/430 servers with 1GB memory. I need to compile DISKD support in FreeBSD kernel. I read in Squid's FAQ which recommends the following values for the message queues and shared memory: if you have FreeBSD5 or higher, you can use aufs, which should be faster and doesn't require SHM tuning. -- Matus UHLAR - fantomas, [EMAIL PROTECTED] ; http://www.fantomas.sk/ Warning: I wish NOT to receive e-mail advertising to this address. Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu. How does cat play with mouse? cat /dev/mouse
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Thanks Matus for the DiskD issue on FreeBSD. One more thing, we have a couple of FreeBSD squid servers. We have both Squid-2.5 and Squid-2.6 running. We do load-balancing among the squid servers by using an Alteon load-balancer. My problem is that I see alot of invalid request in my cache.log from this Alteon load-balancer. Below are the some part of the cache.log details: Please note: x.x.x.x is a real static public IP. On my Server with Squid-2.6 it shows as: clientReadRequest: FD 290 (x.x.x.x:7856) Invalid Request parseHttpRequest: Unsupported method 'aa clientReadRequest: FD 290 (x.x.x.x:7856) Invalid Request parseHttpRequest: Unsupported method 'aa On my Servers with Squid-2.5, it shows as: 2006/11/09 14:28:38| clientReadRequest: FD 119 Invalid Request 2006/11/09 14:28:40| parseHttpRequest: Unsupported method 'aa Only difference with Squid-2.5 and Squid-2.6 is that the real IP of the Alteon gets shown as shown above. In all my squid servers, cache.log shows this type of requests from the Alteon load-balancer once every second. In one day, I can expect to see at least 86400 entries of this type in my cache.log on each of my squid servers. Please shed some light on these issue. On Thu, 9 Nov 2006 09:05:13 +0100 Matus UHLAR - fantomas [EMAIL PROTECTED] wrote: On 08.11.06 22:48, Tek Bahadur Limbu wrote: I am using squid-2.5/2.6 with FreeBSD 6.0/6.1 on some of my Dell SC420/430 servers with 1GB memory. I need to compile DISKD support in FreeBSD kernel. I read in Squid's FAQ which recommends the following values for the message queues and shared memory: if you have FreeBSD5 or higher, you can use aufs, which should be faster and doesn't require SHM tuning. - -- With best regards and good wishes, Yours sincerely, Tek Bahadur Limbu Jwl Systems Department Worldlink Communications Pvt. Ltd. -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.5 (FreeBSD) iD8DBQFFUuw3wXBzLfFnNxQRAovxAJ9JGhXLRUXjKNNh2/TdXJfccNb1fwCgj15G 12BeAbTwsgx5f+7YV/hRuyE= =yzsU -END PGP SIGNATURE-
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
tor 2006-11-09 klockan 14:37 +0545 skrev Tek Bahadur Limbu: clientReadRequest: FD 290 (x.x.x.x:7856) Invalid Request parseHttpRequest: Unsupported method 'aa In all my squid servers, cache.log shows this type of requests from the Alteon load-balancer once every second. In one day, I can expect to see at least 86400 entries of this type in my cache.log on each of my squid servers. Please shed some light on these issue. Look at the traffic with a packet analyzer (i.e. wireshark). The ip:port logged is the key to find the correct data in all captured traffic.. My guess is that your Alteon is sending some crap to your Squids, possibly in some availability probe or similar. Regards Henrik signature.asc Description: Detta är en digitalt signerad meddelandedel
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
At 01:05 AM 11/9/2006, Matus UHLAR - fantomas wrote: On 08.11.06 22:48, Tek Bahadur Limbu wrote: I am using squid-2.5/2.6 with FreeBSD 6.0/6.1 on some of my Dell SC420/430 servers with 1GB memory. I need to compile DISKD support in FreeBSD kernel. I read in Squid's FAQ which recommends the following values for the message queues and shared memory: if you have FreeBSD5 or higher, you can use aufs, which should be faster and doesn't require SHM tuning. FreeBSD 5.x or higher panics when you attempt to use AUFS, UNLESS you apply a very recent patch to the code involving kqueues. (I believe that the patch will be in FreeBSD 6.2-RELEASE but was not in FreeBSD 6.1-RELEASE.) I had to patch the kernel manually to get Squid's AUFS code to work on a client's system without crashing it. The patch can also be applied to 5.x, I'm sure. See http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/103127 --Brett Glass
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
On Thu, Nov 09, 2006, Brett Glass wrote: FreeBSD 5.x or higher panics when you attempt to use AUFS, UNLESS you apply a very recent patch to the code involving kqueues. (I believe that the patch will be in FreeBSD 6.2-RELEASE but was not in FreeBSD 6.1-RELEASE.) I had to patch the kernel manually to get Squid's AUFS code to work on a client's system without crashing it. The patch can also be applied to 5.x, I'm sure. See http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/103127 Does that mean you're successfully running Squid under FreeBSD + AUFS? Adrian
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Thank you very much to everybody for the information. I now know the reason why Squid and the FreeBSD 6.1 box crashes every now and then when I use aufs. I was using Diskd for the past 2 weeks but it caused the Squid process to be zombie when the load goes above 100 requests/second. Also I don't see any kind of useful info on the cache.log when the Squid process gets zombied except for: 2006/11/10 11:28:14| clientReadRequest: FD 142 (x.x.x.x:60171) Invalid Request 2006/11/10 11:28:16| parseHttpRequest: Unsupported method 'aa ' 2006/11/10 11:28:16| clientReadRequest: FD 253 (x.x.x.x:60181) Invalid Request 2006/11/10 11:28:18| parseHttpRequest: Unsupported method 'aa ' 2006/11/10 11:28:18| clientReadRequest: FD 96 (x.x.x.x:60191) Invalid Request 2006/11/10 11:28:20| parseHttpRequest: Unsupported method 'aa I will apply the patch for aufs in my FreeBSD 6.1 (amd 64 bit) server tonight. For the time being, I will try to use ufs. By the way guys, I have these compiled in my FreeBSD 6.1 kernel: options IPFIREWALL #firewall options IPFIREWALL_VERBOSE #enable logging to syslogd(8) options IPFIREWALL_FORWARD #enable transparent proxy options IPFIREWALL_VERBOSE_LIMIT=1000#limit verbosity options IPFIREWALL_FORWARD_EXTENDED #all packet dest changes #DiskD Shared Memory and Message Queue options MSGMAX=32768 (max characters in a message) options MSGMNB=16384 options MSGMNI=41 options MSGSEG=2049 options MSGSSZ=64 options MSGTQL=2048 options SHMSEG=128 options SHMMNI=192 options SHMMAX=33554432 options SHMMIN=1 options SHMALL=8192 options MAXDSIZ=2147483648 options DFLDSIZ=536870912 options CPU_ENABLE_SSE options UFS_DIRHASH options KVA_PAGES=512 #other settings (requires a lot of RAM) options MAXFILES=8192 options NMBCLUSTERS=32768 Please shed some light on this. On Fri, 10 Nov 2006 13:05:06 +0800 Adrian Chadd [EMAIL PROTECTED] wrote: On Thu, Nov 09, 2006, Brett Glass wrote: FreeBSD 5.x or higher panics when you attempt to use AUFS, UNLESS you apply a very recent patch to the code involving kqueues. (I believe that the patch will be in FreeBSD 6.2-RELEASE but was not in FreeBSD 6.1-RELEASE.) I had to patch the kernel manually to get Squid's AUFS code to work on a client's system without crashing it. The patch can also be applied to 5.x, I'm sure. See http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/103127 Does that mean you're successfully running Squid under FreeBSD + AUFS? Adrian - -- With best regards and good wishes, Yours sincerely, Tek Bahadur Limbu Jwl Systems Department Worldlink Communications Pvt. Ltd. -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.5 (FreeBSD) iD8DBQFFVBRhwXBzLfFnNxQRAsCYAKCdDNvSHDbbPXnboTO+/CmFbaPWtgCg4MZZ vh7ctwUqYe1rM65b5oFNIUE= =w0mu -END PGP SIGNATURE-
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
On Fri, Nov 10, 2006, Tek Bahadur Limbu wrote: Please shed some light on this. Diskd has some usage cases which could cause crashes. We thought about fixing them but then realised: * the only OS where diskd makes more sense is FreeBSD * and FreeBSD now has a sane thread model; so it can use AUFS So the suggestion from us is upgrade to the latest freebsd-6; apply the patch that brett suggested and use aufs. As I said to someone else a few weeks ago - we could fix diskd but it'd require quite a bit of effort and there's other things higher on the priority list for us to fix. If you'd like diskd to be fixed then please let us know. But please realise a lot of this work is stuff we're doing for fun now. Not a lot of us are actually paid to work on Squid for any appreciable amount of time - the only Squid developer I know who derives income from Squid is Henrik. I do this stuff for fun. Adrian
Re: [squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Hello Alexandre and Adrian, Actually, these are the requests of my Alteon load-balancer which is used to divide the load among our dozen Squid proxy servers. But I don't know why the Alteon load-balancer are sending those kind of requests. Adrian, I really want to thank you for your response on Diskd. I will definitely go for AUFS rather than Diskd. However please note that we are only using FreeBSD servers to run squid. I have used the following options while compiling Squid-2.6 on one of my FreeBSD 6.1 server. ./configure '--bindir=/usr/local/sbin' '--sysconfdir=/usr/local/etc/squid' '--datadir=/usr/local/etc/squid' '--libexecdir=/usr/local/libexec/squid' '--localstatedir=/usr/local/squid' '--enable-removal-policies=lru heap' '--enable-async-io' '--with-pthreads' '--enable-storeio=ufs diskd null aufs' '--enable-delay-pools' '--enable-snmp' '--enable-carp' '--enable-htcp' '--enable-auth=ntlm,basic' '--enable-cache-digests' '--enable-underscores' '--enable-useragent-log' '--enable-async-io=48' '--enable-time-hack' '--enable-arp-acl' '--enable-kqueue' '--enable-poll' '--enable-select' '--enable-icp' '--enable-pf-transparent' '--enable-ipf-transparent' '--with-large-files' '--enable-large-cache-files' '--enable-follow-x-forwarded-for' '--enable-ssl' '--prefix=/usr/local' 'amd64-portbld-freebsd6.1' 'build_alias=amd64-portbld-freebsd6.1' 'host_alias=amd64-portbld-freebsd6.1' 'target_alias=amd64-portbld-freebsd6.1' Are there any errors and problems in my compilation? Any help and suggestions will really be appreciated. On Fri, 10 Nov 2006 03:09:48 -0300 Alexandre Correa [EMAIL PROTECTED] wrote: i received this message too... i think this is some P2P connections on users using port 80 (eg: emule configurated on port 80 to bypass protections) ... On 11/10/06, Tek Bahadur Limbu [EMAIL PROTECTED] wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Thank you very much to everybody for the information. I now know the reason why Squid and the FreeBSD 6.1 box crashes every now and then when I use aufs. I was using Diskd for the past 2 weeks but it caused the Squid process to be zombie when the load goes above 100 requests/second. Also I don't see any kind of useful info on the cache.log when the Squid process gets zombied except for: 2006/11/10 11:28:14| clientReadRequest: FD 142 Invalid Request 2006/11/10 11:28:16| parseHttpRequest: Unsupported method 'aa ' 2006/11/10 11:28:16| clientReadRequest: FD 253 (x.x.x.x:60181) Invalid Request 2006/11/10 11:28:18| parseHttpRequest: Unsupported method 'aa ' 2006/11/10 11:28:18| clientReadRequest: FD 96 (x.x.x.x:60191) Invalid Request 2006/11/10 11:28:20| parseHttpRequest: Unsupported method 'aa I will apply the patch for aufs in my FreeBSD 6.1 (amd 64 bit) server tonight. For the time being, I will try to use ufs. By the way guys, I have these compiled in my FreeBSD 6.1 kernel: options IPFIREWALL #firewall options IPFIREWALL_VERBOSE #enable logging to syslogd(8) options IPFIREWALL_FORWARD #enable transparent proxy options IPFIREWALL_VERBOSE_LIMIT=1000 #limit verbosity options IPFIREWALL_FORWARD_EXTENDED #all packet dest changes #DiskD Shared Memory and Message Queue options MSGMAX=32768 (max characters in a message) options MSGMNB=16384 options MSGMNI=41 options MSGSEG=2049 options MSGSSZ=64 options MSGTQL=2048 options SHMSEG=128 options SHMMNI=192 options SHMMAX=33554432 options SHMMIN=1 options SHMALL=8192 options MAXDSIZ=2147483648 options DFLDSIZ=536870912 options CPU_ENABLE_SSE options UFS_DIRHASH options KVA_PAGES=512 #other settings (requires a lot of RAM) options MAXFILES=8192 options NMBCLUSTERS=32768 Please shed some light on this. On Fri, 10 Nov 2006 13:05:06 +0800 Adrian Chadd [EMAIL PROTECTED] wrote: On Thu, Nov 09, 2006, Brett Glass wrote: FreeBSD 5.x or higher panics when you attempt to use AUFS, UNLESS you apply a very recent patch to the code involving kqueues. (I believe that the patch will be in FreeBSD 6.2-RELEASE but was not in FreeBSD 6.1-RELEASE.) I had to patch the kernel manually to get Squid's AUFS code to work on a client's system without crashing it. The patch can also be applied to 5.x, I'm sure. See http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/103127 Does that mean you're successfully running Squid under FreeBSD + AUFS? Adrian - -- With best regards and good wishes, Yours sincerely, Tek Bahadur Limbu Jwl Systems Department Worldlink Communications Pvt. Ltd. -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.5 (FreeBSD)
[squid-users] Diskd Message Queues and Shared memory for FreeBSD 6.x
Dear All, I am using squid-2.5/2.6 with FreeBSD 6.0/6.1 on some of my Dell SC420/430 servers with 1GB memory. I need to compile DISKD support in FreeBSD kernel. I read in Squid's FAQ which recommends the following values for the message queues and shared memory: Message Queues: options MSGMNB=8192 # max # of bytes in a queue options MSGMNI=40 # number of message queue identifiers options MSGSEG=512 # number of message segments per queue options MSGSSZ=64 # size of a message segment options MSGTQL=2048 # max messages in system options SHMSEG=16 # max shared mem id's per process options SHMMNI=32 # max shared mem id's per system options SHMMAX=2097152 # max shared memory segment size (bytes) options SHMALL=4096 # max amount of shared memory (pages Shared Memory: Now, I guess that these values were recommended for a system built at least 3/4 years ago. My question is Are the above values good enough for recent versions of FreeBSD such as 6.0 or 6.1? -- ... With best regards and good wishes, Yours sincerely, Tek Bahadur Limbu Jwl Systems Department Worldlink Communications signature.asc Description: This is a digitally signed message part
RE: [squid-users] diskd
I am using the squid originally installed with my Fedora Club 3. So I don't know if diskd is active. I am interested on it since I read squid FAQ's recomending it. Use cachemgr - Current Squid Configuration. Check the store method used for the define cache dirs. On Linux aufs behaves better then diskd. M.
RE: [squid-users] diskd
So I won´t worry anymore, will keep aufs. :) Daniel Navarro Maracay, Venezuela www.csaragua.com/ecodiver --- Elsen Marc [EMAIL PROTECTED] escribió: I am using the squid originally installed with my Fedora Club 3. So I don't know if diskd is active. I am interested on it since I read squid FAQ's recomending it. Use cachemgr - Current Squid Configuration. Check the store method used for the define cache dirs. On Linux aufs behaves better then diskd. M. _ Do You Yahoo!? Información de Estados Unidos y América Latina, en Yahoo! Noticias. Visítanos en http://noticias.espanol.yahoo.com
RE: [squid-users] diskd
--- Elsen Marc [EMAIL PROTECTED] escribió: How can I implement and test diskd? I am using the squid integrated into Fedora Club 3. - Use the appropriate configure option to select this store method - Select this method when configuring cache_dirs in squid.conf. Check squid.conf.default for explanations and comments. M. Thanks for replying back. I am using the squid originally installed with my Fedora Club 3. So I don't know if diskd is active. I am interested on it since I read squid FAQ's recomending it. Regards, Daniel Navarro Maracay, Venezuela www.csaragua.com/ecodiver _ Do You Yahoo!? Información de Estados Unidos y América Latina, en Yahoo! Noticias. Visítanos en http://noticias.espanol.yahoo.com
Re: [squid-users] diskd
On Sat, 2005-01-22 at 20:57 -0600, Daniel Navarro wrote: Hi all How can I implement and test diskd? Just configure a cache_dir of type diskd. please refer to the squid.conf.default you can find in /etc/squid for the specific parameters. Kinkie
RE: [squid-users] diskd
How can I implement and test diskd? I am using the squid integrated into Fedora Club 3. - Use the appropriate configure option to select this store method - Select this method when configuring cache_dirs in squid.conf. Check squid.conf.default for explanations and comments. M.
[squid-users] diskd
Hi all How can I implement and test diskd? I am using the squid integrated into Fedora Club 3. Cheers, Daniel Navarro Maracay, Venezuela www.csaragua.com/ecodiver _ Do You Yahoo!? Información de Estados Unidos y América Latina, en Yahoo! Noticias. Visítanos en http://noticias.espanol.yahoo.com
[squid-users] DiskD
hi all i m using squid-2.5.stable1-2.i385.rpm came with red hat linux 9 do i need to recompile it to use diskd or aufs Regards Azeem _ Tired of spam? Get advanced junk mail protection with MSN 8. http://join.msn.com/?page=features/junkmail
Re: [squid-users] DiskD
On Tue, 5 Oct 2004, azeem ahmad wrote: i m using squid-2.5.stable1-2.i385.rpm came with red hat linux 9 do i need to recompile it to use diskd or aufs /usr/sbin/squid -v will tell you what options your Squid binary is built with. Trying to use the features and verifying with /usr/sbin/squid -k parse is also a good way. Regards Henrik
Re: [squid-users] diskd or coss?
On 20.08 13:01, Costas Zacharopoulos wrote: I have slackware linux and a ext3fs partition for squid cache. Which cache type has better performance ? changing fs would help you much more for the performance. try reisersfs or xfs, if you find it possible -- Matus UHLAR - fantomas, [EMAIL PROTECTED] ; http://www.fantomas.sk/ Warning: I wish NOT to receive e-mail advertising to this address. Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu. Windows found: (R)emove, (E)rase, (D)elete
[squid-users] diskd or coss?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 I have slackware linux and a ext3fs partition for squid cache. Which cache type has better performance ? -BEGIN PGP SIGNATURE- Version: GnuPG v1.2.3 (GNU/Linux) iD8DBQFBJcwJbKZ5WTJlMpgRArcdAJ0fXMjjVLauPIzV6HyeWWiUlqmRIQCgurvH 5hO8+YMwwf4oJleKVbmHe4Q= =OCsu -END PGP SIGNATURE-
Re: [squid-users] diskd or coss?
On Fri, 20 Aug 2004, Costas Zacharopoulos wrote: I have slackware linux and a ext3fs partition for squid cache. Which cache type has better performance ? For Linux you should be using the aufs cache_dir type. diskd is mainly for FreeBSD and alikes. coss is still under development and not yet very useable. Regards Henrik
Re: [squid-users] Diskd settings questions
Anyone have any input on this? It would benefit me as well. Thanks. -gvb On Jul 2, 2004, at 6:18 PM, Pete Davis wrote: I have successfully set up several Squid boxes on FreeBSD (4.x and 5.x). They are running well with no problems to speak of at the time. Searching through the mailing list archives does wonders for your knowledge. I do have questions about Diskd settings: 1)what are the optimum settings for Diskd on FreeBSD? The FAQ has information on the settings but warns, You can set the parameters in the kernel as follows. This is just an example. Make sure the values are appropriate for your system:. I cannot find how to adjust the values based on the box (I have several different sized servers). I use the default settings but I have never really stressed the box enough to know if those settings (from the FAQ) are good if/when the load goes up. I am one of the I like to know before I have to know (before a problem??) kind of people. 2) Should I change to the settings in the FAQ even though many of those settings are much less than what the current settings are (GENERIC kernel). Latest server (this is a test box to do testing on every possible combination I might need- transparent, null cache, diskd, aufs, different replacement options, different authorization schemes, etc) is: PIII 933, 512mb, 2x9scsi on FreeBSD 5.2.1. This box has the following settings for SYSVSHM (current settings followed by FAQ suggestion): SHMSEG=128 (FAQ= 16) SHMMNI=192 (FAQ=32) SHMMAX=33554432 (FAQ=2097152) SHMALL=8194 (FAQ=4096) I'll eventually blow the config away and load SuSE Enterprise server on it (then learn how to optimize a Linux server). There seems to be more information about settings for Squid+Linux. Thanks for any light you may be able to shed on this.
[squid-users] Diskd settings questions
I have successfully set up several Squid boxes on FreeBSD (4.x and 5.x). They are running well with no problems to speak of at the time. Searching through the mailing list archives does wonders for your knowledge. I do have questions about Diskd settings: 1)what are the optimum settings for Diskd on FreeBSD? The FAQ has information on the settings but warns, You can set the parameters in the kernel as follows. This is just an example. Make sure the values are appropriate for your system:. I cannot find how to adjust the values based on the box (I have several different sized servers). I use the default settings but I have never really stressed the box enough to know if those settings (from the FAQ) are good if/when the load goes up. I am one of the I like to know before I have to know (before a problem??) kind of people. 2) Should I change to the settings in the FAQ even though many of those settings are much less than what the current settings are (GENERIC kernel). Latest server (this is a test box to do testing on every possible combination I might need- transparent, null cache, diskd, aufs, different replacement options, different authorization schemes, etc) is: PIII 933, 512mb, 2x9scsi on FreeBSD 5.2.1. This box has the following settings for SYSVSHM (current settings followed by FAQ suggestion): SHMSEG=128 (FAQ= 16) SHMMNI=192 (FAQ=32) SHMMAX=33554432 (FAQ=2097152) SHMALL=8194 (FAQ=4096) I'll eventually blow the config away and load SuSE Enterprise server on it (then learn how to optimize a Linux server). There seems to be more information about settings for Squid+Linux. Thanks for any light you may be able to shed on this.
RE: [squid-users] diskd VS aufs
Hi all, I've read that aufs and diskd are much much better than ufs. I am trying to decide between the two. I've read that aufs is the better one (in terms of performance), but probabability of crashing is higher? I haven't seen problems with my aufs based configuration in months I am willing to sacrifice a bit of performance gain but would like to have maximum stability. Can anyone advice which would be the better choice? The choice is OS dependend : under FreeBSD diskd is the better choice. Under Solaris and Linux use aufs. M.
RE: [squid-users] diskd VS aufs
Thanks. I suppose I do not need to rebuild the cache (./squid -z). But what makes you conclude that FreeBSD diskd is the better choice. Under Solaris and Linux use aufs.? Regards Liz From: Elsen Marc [EMAIL PROTECTED] To: Lizzy Dizzy [EMAIL PROTECTED],[EMAIL PROTECTED] Subject: RE: [squid-users] diskd VS aufs Date: Wed, 30 Jun 2004 08:00:43 +0200 Hi all, I've read that aufs and diskd are much much better than ufs. I am trying to decide between the two. I've read that aufs is the better one (in terms of performance), but probabability of crashing is higher? I haven't seen problems with my aufs based configuration in months I am willing to sacrifice a bit of performance gain but would like to have maximum stability. Can anyone advice which would be the better choice? The choice is OS dependend : under FreeBSD diskd is the better choice. Under Solaris and Linux use aufs. M. _ Find love on MSN Personals http://personals.msn.com.sg/
RE: [squid-users] diskd VS aufs
Thanks. I suppose I do not need to rebuild the cache (./squid -z). But what No, you don't. makes you conclude that FreeBSD diskd is the better choice. Under Solaris and Linux use aufs.? I seem to remember this was a FAQ, I may be wrong, at least I can't find it for the moment. From what I remember : under FreeBSD support for shared mem is better which diskd heavily uses. Linux and Solaris are better on thread support, which aufs uses. M.
RE: [squid-users] diskd VS aufs
On Wed, 30 Jun 2004, Elsen Marc wrote: Thanks. I suppose I do not need to rebuild the cache (./squid -z). But what No, you don't. makes you conclude that FreeBSD diskd is the better choice. Under Solaris and Linux use aufs.? I seem to remember this was a FAQ, I may be wrong, at least I can't find it for the moment. From what I remember : under FreeBSD support for shared mem is better which diskd heavily uses. Linux and Solaris are better on thread support, which aufs uses. M. We are running Squid on FreeBSD 5.2-CURRENT with aufs. It looks that KSE improved the thread support in the FreeBSD. Val
[squid-users] diskd VS aufs
Hi all, I've read that aufs and diskd are much much better than ufs. I am trying to decide between the two. I've read that aufs is the better one (in terms of performance), but probabability of crashing is higher? I am willing to sacrifice a bit of performance gain but would like to have maximum stability. Can anyone advice which would be the better choice? Thanks Liz _ Find it on the web with MSN Search. http://search.msn.com.sg/
[squid-users] diskd: number of processes
According to the FAQ docs about diskd, (http://www.squid-cache.org/Doc/FAQ/FAQ-22.html) each cache_dir has its own diskd child process. Does it mean that...if i modify the squid.conf like below : cache_dir diskd /reiser/squid-cache-diskd/1 10240 32 512 cache_dir diskd /reiser/squid-cache-diskd/2 10240 32 512 cache_dir diskd /reiser/squid-cache-diskd/3 10240 32 512 , and then execute Squid, are there 3 child processes of Squid like below? (It's just an example) ps -ef | grep squid root 7123 1 0 18:14 ?00:00:00 sbin/squid nobody7125 7123 0 18:14 ?00:00:03 (squid) nobody7125 7123 0 18:14 ?00:00:03 (squid) nobody7125 7123 0 18:14 ?00:00:03 (squid) root 7440 5505 0 18:28 pts/000:00:00 grep squid Do I misunderstand about diskd? Thanks in advance. _ MSN Messenger . http://messenger.msn.co.kr
Re: [squid-users] diskd: number of processes
On Thu, 19 Feb 2004, Ko Jong Hyun wrote: According to the FAQ docs about diskd, (http://www.squid-cache.org/Doc/FAQ/FAQ-22.html) each cache_dir has its own diskd child process. Yes. Does it mean that...if i modify the squid.conf like below : cache_dir diskd /reiser/squid-cache-diskd/1 10240 32 512 cache_dir diskd /reiser/squid-cache-diskd/2 10240 32 512 cache_dir diskd /reiser/squid-cache-diskd/3 10240 32 512 , and then execute Squid, are there 3 child processes of Squid like below? (It's just an example) ps -ef | grep squid root 7123 1 0 18:14 ?00:00:00 sbin/squid nobody7125 7123 0 18:14 ?00:00:03 (squid) nobody7125 7123 0 18:14 ?00:00:03 (squid) nobody7125 7123 0 18:14 ?00:00:03 (squid) root 7440 5505 0 18:28 pts/000:00:00 grep squid Close, but not quite correct. It will look more like root 7123 1 0 18:14 ?00:00:00 sbin/squid nobody7125 7123 0 18:14 ?00:00:03 (squid) nobody7126 7125 0 18:14 ?00:00:03 diskd nobody7127 7125 0 18:14 ?00:00:03 diskd nobody7127 7125 0 18:14 ?00:00:03 diskd There may also be other slave processes to the main Squid process (Pid 7125 in your example), depending on what other features of Squid you use. Regards Henrik
Re: [squid-users] diskd - option
Hello Henrik, Thanks for the reply. I've seen Releasenotes. As i said in prev mail, samba version is 2.2.6 But i came to know that winbind helpers updated to match Samba-2.2.7a and should work with Samba-2.2.6 or later (required). So my pbl is runtime pbl or build pbl? i mean will it(wb_auth or wb_group) run by chnaging samba verion (it is build where the machine has samba-2.2.3a.) or i need to rebuild squid after having samba 2.2.6 verison. plz help me. Regs, -Sadha --- Henrik Nordstrom [EMAIL PROTECTED] wrote: On Mon, 1 Dec 2003, shadha nker wrote: ***my samba verison is 2.2.3a and 2.2.5. I've one dbt then how with this version , squid2.5STABLE1, wb_auth and wb_group works, but in squid2.5STABLE4 WON't. See the Squid release notes. Regards Henrik __ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/
Re: [squid-users] diskd - option
On Tue, 2 Dec 2003, shadha nker wrote: Thanks for the reply. I've seen Releasenotes. As i said in prev mail, samba version is 2.2.6 But i came to know that winbind helpers updated to match Samba-2.2.7a and should work with Samba-2.2.6 or later (required). So my pbl is runtime pbl or build pbl? A build problem. i mean will it(wb_auth or wb_group) run by chnaging samba verion (it is build where the machine has samba-2.2.3a.) or i need to rebuild squid after having samba 2.2.6 verison. You either need to change Samba version or rebuild the helpers to use your older Samba version according to the instructions in the release notes. Regards Henrik
Re: [squid-users] diskd - option
Hello Henrik, Thanks for your response. So one solution is I can change samba version = 2.2.6 to run this itself and no need to rebuild for newer samba Thanks . If anythong wrong in my above statement,plz reply. Regs, -Sadha --- Henrik Nordstrom [EMAIL PROTECTED] wrote: On Tue, 2 Dec 2003, shadha nker wrote: Thanks for the reply. I've seen Releasenotes. As i said in prev mail, samba version is 2.2.6 But i came to know that winbind helpers updated to match Samba-2.2.7a and should work with Samba-2.2.6 or later (required). So my pbl is runtime pbl or build pbl? A build problem. i mean will it(wb_auth or wb_group) run by chnaging samba verion (it is build where the machine has samba-2.2.3a.) or i need to rebuild squid after having samba 2.2.6 verison. You either need to change Samba version or rebuild the helpers to use your older Samba version according to the instructions in the release notes. Regards Henrik __ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/
Re: [squid-users] diskd - option
On Sun, 30 Nov 2003, shadha nker wrote: My Q! is there a way to run ./libexec/diskd with any command line option(if so how? one example plz) or there is a only with diskd_program with squid.conf? diskd is not usable from the command line. It must be called via squid.conf. diskd does not take any command line arguments. It gets what it needs to know from Squid. Why are you looking for command line arguments to diskd? Regards Henrik
Re: [squid-users] diskd - option
Hello Henrik, Thankyou very much for your reply. After starting squid, squid is running properly. when try to put ps -ae|grep diskd, diskd process is also running. But when i try to run as ./libexec/diskd Assertion failed: 4 == argc, file diskd.c, line 289 Abort(coredump) I don't know whats the reason for coredump. So only i doubted that, May be i'm worng. Could u plz give me the reason for coredump? Or it is not possible to run as above. Then same way when i try to run wb_auth and wb_group, it gives as # ./wb_auth /wb_auth[23669](wb_basic_auth.c:161): Can't contact winbindd. Dying # ./wb_group /wb_group[23670](wb_check_group.c:309): Can't contact winbindd. Dying I'm using squid-2.5STABLE4. But squid-2.5STABLE1 works fine. plz help to resolve this. Advance thanks for your help. Regs, sadha --- Henrik Nordstrom [EMAIL PROTECTED] wrote: On Sun, 30 Nov 2003, shadha nker wrote: My Q! is there a way to run ./libexec/diskd with any command line option(if so how? one example plz) or there is a only with diskd_program with squid.conf? diskd is not usable from the command line. It must be called via squid.conf. diskd does not take any command line arguments. It gets what it needs to know from Squid. Why are you looking for command line arguments to diskd? Regards Henrik __ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/
Re: [squid-users] diskd - option
On Mon, 1 Dec 2003, shadha nker wrote: But when i try to run as ./libexec/diskd Assertion failed: 4 == argc, file diskd.c, line 289 Abort(coredump) I don't know whats the reason for coredump. You are not supposed to execute diskd this way, and it most likely will crash and burn if you do. We could add code to diskd detecting if someone tries to execute it manaually and give an error message, but in the 3 years diskd has been part of the distribution you are the only one who has taken up this discussion. Then same way when i try to run wb_auth and wb_group, it gives as # ./wb_auth /wb_auth[23669](wb_basic_auth.c:161): Can't contact winbindd. Dying # ./wb_group /wb_group[23670](wb_check_group.c:309): Can't contact winbindd. Dying Most likely you do not have winbindd running on your system. Or you have another version of Samba than the helpers was designed for. Regards Henrik
[squid-users] Diskd status question
Hello Everyone, Can some explain what is open_fail_queue_len and block_queue_len means here. Since there is no FAIL Can I conclude that everything is working properly? Regards, Nooshin DISKD EVENT STATUS. sent_count: 27676065 recv_count: 27676065 max_away: 73 max_shmuse: 73 open_fail_queue_len: 48776 block_queue_len: 1125821217 OPS SUCCESSFAIL open 3965906 3965906 0 create 1652376 1652376 0 close 5618280 5618280 0 unlink 1474312 1474312 0 read 8111483 8111483 0 write 6853708 6853708 0
[squid-users] diskd error messages... :(
dear all... why my diskd showed the error message like this : Jul 18 17:04:51 calculus squid[22823]: ctx: enter level 0: 'http://202.59.204.27/mrtg/202.59.204.27_21-day.png' Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend: msgsnd: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend UNLINK: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: ctx: exit level 0 Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend: msgsnd: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend OPEN: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend: msgsnd: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend OPEN: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend: msgsnd: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend OPEN: (35) Resource temporarily unavailable can anyone tell me what is wrong? regards, tommy chandra wiratama
Re: [squid-users] diskd error messages... :(
fre 2003-07-18 klockan 11.59 skrev Tommy C Wiratama: dear all... why my diskd showed the error message like this : Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend: msgsnd: (35) Resource temporarily unavailable Jul 18 17:04:51 calculus squid[22823]: storeDiskdSend OPEN: (35) Resource temporarily unavailable can anyone tell me what is wrong? See the Squid FAQ on diskd configuration. Your OS is not configured with room for enough messages for diskd. Regards Henrik -- Donations welcome if you consider my Free Squid support helpful. https://www.paypal.com/xclick/business=hno%40squid-cache.org Please consult the Squid FAQ and other available documentation before asking Squid questions, and use the squid-users mailing-list when no answer can be found. Private support questions is only answered for a fee or as part of a commercial Squid support contract. If you need commercial Squid support or cost effective Squid and firewall appliances please refer to MARA Systems AB, Sweden http://www.marasystems.com/, [EMAIL PROTECTED]