Re: [squid-users] Linux 2.6 Kernel. Any benefits?
I'm running smoothly a RH 9 server with kernel 2.6.6 and squid 2.5 stable 4, with great improvements on VM handling , before kernel 2.6 it use to swap even thou I don't have a memory shortage . []'s On Wed, 23 Jun 2004 06:11:25 +, Lizzy Dizzy [EMAIL PROTECTED] wrote: Hi All, I am keen to switch my kernel to 2.6.7, due to reports that it is much more efficient that the current 2.4 kernel. I am currently using Squid 2.5-S4. Has anybody switched from a 2.4 kernel to a 2.6 kernel? Do you really see a performance boost in squid? Thanks Liz _ Take a break! Find destinations on MSN Travel. http://www.msn.com.sg/travel/ -- If you really want something in this life, you have to work for it. Now, quiet! They're about to announce the lottery numbers... - Homer Simpson
Re: [squid-users] How often should I restart Squid?
On Mon, 28 Jun 2004 14:05:01 +0200, Matus UHLAR - fantomas [EMAIL PROTECTED] wrote: On 25.06 22:19, Bruno Marcondes wrote: I had this weird behavior with squid on a RH9 server using kernel 2.4.22 , 2Gb of memory (the server was DL380 G3 from Compaq/HP), the swap usage was slowly growing forever and I even try disabling it with no harms , but I didn't want do it ...why ? It shouldn't swap ! it did not swap. It is just linux kernel behaviour that it saves some memory pages onto swap even if there's no need for it. IT is an optimalization feature so that makes nothing bad for you. If you have enough memory, you really can turn off swap. But it was consuming swap resources without ever freeing it. I take care of a lot of other linux boxes running apache+mod_mem_cache , apache + jrun/jetty + java and other memory consuming applications and have never seen this swap indefinitely growing usage behavior on then. After all kernel-2.6.6 gave my squid server more performance and that behavior is gone.
Re: [squid-users] How often should I restart Squid?
Folks, I had this weird behavior with squid on a RH9 server using kernel 2.4.22 , 2Gb of memory (the server was DL380 G3 from Compaq/HP), the swap usage was slowly growing forever and I even try disabling it with no harms , but I didn't want do it ...why ? It shouldn't swap ! My squid is set behind a apache + mod_proxy as a web accelerator caching requests for other sites in a reverse proxy configuration. It is very busy ( ~ 160 reqs/s ) . 10 days ago I upgrade to kernel 2.6.6, guess what . NO swap usage anymore ! All the memory is allocated, but ~50 % of it is cached . Upgrading the kernel on my squid servers went pretty good for me so far ... regards Bruno Marcondes
Re: [squid-users] How often should I restart Squid?
On Mon, 21 Jun 2004 03:31:26 +, Lizzy Dizzy [EMAIL PROTECTED] wrote: When squid started using so much swapspace, does it means that there's many obsolete objects inside the RAM that has been forgotten and thus will not be removed/swapped out? What is your memory setting on squid.conf ? Kernel version ?
Re: [squid-users] Info about files and performance
Diego Ariel Dasso wrote: Hello, is there a relation in performance between the size of squid.conf, porn.block, porn.desblock and squid? i.e. by eliminatig all comentaries in squid.conf the performance is better or by having a large list in porn.block the performance is worst? Coments dont make no difference for performance. The number of entries on your block list will probaply only decrease performance when it reachs some milions of lines. So I guess that if you are heaving performance problems, this is not the reason . Bruno
Re: [squid-users] To Have A /swap Partition, Or Not?
I have a strange situation, where I wonder if disabling /swap will help, or solve. My squid server runs as a http accelerator, it is a HP/Compaq DL360 with 2 Gb of memory, 4 squid process running as it has 2 hyperthread cpus. It serves ~ 150 req/s (each process), its Linux RH9 kernel 2.4. After some time running/serving , memory usage looks like this : total used free sharedbuffers cached Mem: 20694882059504 9984 0 721641466184 -/+ buffers/cache: 5211561548332 Swap: 2097112 1191281977984 The thing is that swap usage keeps growing until the machine consumes all the 2G swap partition. It took weeks to fill up the swap area, but eventually it will be 100%. I set squid cache_mem to 32 mb because i guessed that this 1.5 cached memory could only be file cache, as this machine does nothing more than running squid. 14781 nobody18 0 105M 100M 3180 NYI S27.1 4.9 1802m squid 8 nobody12 0 101M 100M 2192 NYI S 3.8 4.9 1333m squid 22358 nobody16 0 69148 66M 8248 NYI S19.4 3.3 399:07 squid 22374 nobody13 0 69488 66M 3344 NYI S 7.7 3.2 414:42 squid So each squid process takes no more than 100mb, and I have some redir perl scripts running, but only takes like 50 mb more. Afterall , this is very odd . I'm looking forward for any sugestions . Thanks in advance. Regards Bruno Marcondes
[squid-users] Linux eating all memory
Hi folks, I'm having a problem of memory shortage on my squid server. The server is an Compaq DL350 G3 with 2G of ram with linux RH7 / kern. 2.4.22 . It doesnt matter how I set cache_mem , kernel keeps caching (file cache?) till it uses all memory , leaving only 10Mb, thus the server keeps swapping all the time ! I search google for ways of setting vm on 2.4 to *not* eat all my memory, but doesnt find nothing really useful till now ! Thanks in advance ! Bruno Marcondes []'s
Re: [squid-users] Async I/O with RH 9 (NPTL) problem ?
Hi Jason, In fact, the way out for me was to change to diskd , even using RH 7.3 (no nptl) I still had that problem . My reverse proxy has a frontend with apache+mod_proxy+mod_rewrite, and squid on backend for all traffic, so its very loaded. I tested my disks and they wasnt the limitation, so I guess that it is an AsyncIO thread code performance limit on linux plataform (I didn't get this behavior on Solaris). On my server (HP/Compaq DL380 G3) diskd seens to be doing the work ! []'s - Original Message - From: Jason Pickering [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, November 04, 2003 12:27 AM Subject: [squid-users] Async I/O with RH 9 (NPTL) problem ? Hi Bruno, I am experiencing very similar problems with my transparent proxy on a very similar system. Did you ever manage to resolve this problem? Thanks for any tips! Best regards, Jason Pickering
[squid-users] Highlander squid (there can only be one?!)
Hi everyone ! I want to run to squid process on the same machine (as an httpd accellerator) using diferrent confs , thus cache_dirs,logs,pid file ,etc ... I hacked RunAccel run a different squid command (with an alteranate config file), but when the second squid starts it dies with the message : 2003/10/23 19:42:50| Squid is already running! Process ID 25218 Any ideas ? Does anyone already set something like that ? Thanx in advance !
[squid-users] Async I/O with RH 9 (NPTL) problem ?
Hi everyone ! I'm testing squid-2.5STABLE4 on a compaq ProLiant DL380 G3 (2 Xeon 2.4 with hiperthread) and 2 Gb memory running Red Hat 9 . My compile options was : configure options: --host=i386-redhat-linux --with-aufs-threads=32 --build=i386-redhat-linux --target=i386-redhat-linux-gnu --enable-removal-policies=heap,lru --with-pthreads --enable-storeio=aufs,coss,diskd,ufs --enable-snmp --disable-hostname-checks --enable-underscores --disable-ident-lookups --prefix=/opt/squid-2.5.STABLE4 I'm using squid as a transparent proxy (http accelerator) and I set my cache_dir as aufs (with 32 threads set on compile) on ext2 partition that is mounting a 36Gb ultra-scsi disk. Nothing access this disk except squid. Now the problem is : squidaio_queue_request: WARNING - Queue congestion I'm getting this messages a lot on cache.log and after some time squid eventually freezes after showing the following messages : 2003/09/25 17:56:26| squidaio_queue_request: WARNING - Queue congestion 2003/09/25 11:57:43| ctx: enter level 0: 'http://humortadela1.uol.com.br/h/piad/im/pia f' 2003/09/25 11:57:43| squidaio_queue_request: WARNING - Disk I/O overloading 2003/09/25 11:57:43| squidaio_queue_request: Queue Length: current=787, high=787, low=3 2003/09/25 11:57:59| ctx: exit level 0 2003/09/25 11:57:59| ctx: enter level 0: 'http://humortadela1.uol.com.br/h/piad/im/piaf' 2003/09/25 11:57:59|squidaio_queue_request: Async request queue growing uncontrollably! 2003/09/25 11:57:59|squidaio_queue_request: Syncing pending I/O operations.. (blocking) I benchmark the disk with IOzone and got 50Mb/s of throughput! I didn't have this problem at all (not even the warnings) when using Red Hat 7.3. So I'm wondering if squid Async I/O have problems using NPTL (native posix thread library) implemented within RH9 I'm now trying to disable NPTL on RH9. I found out that exporting LD_ASSUME_KERNEL=2.2.5 should disable it , but I didn't overcome the problem doing it on squid compilation and putting on RunAccel script. I'm almost downgrading my linux installation to RH 7.3 now ... Does anyone have successfully run squid-2.5STABLE4 on RH9 that handles good traffic on it (like 80 Mbs) ? Thanx everyone. Bruno Marcondes UOL Sysadmin. www.uol.com.br