[Gluster-users] Gluster 2.0.3 + Apache on CentOS5 performance issue
Somsak Sriprayoonsakul
somsaks at gmail.com
Mon Aug 3 10:59:20 UTC 2009
I have run html read benchmark test using exactly the same old testset.
- Runing normal httpd (prefork) over NFS-sync option - Average TPS ~ 1500,
Peak TPS ~ 3500
- Running normal httpd (prefork) using booster - Average TPS < 500, Peak TPS
~ 1200, slight improvements but each HTTPD still eat up about 50MB+ (keep
increasing) memory instead of about 20MB without booster. I have to cut the
number of httpd process by half, but still the process used up some swap
space.
- Running httpd.worker using booster - The benchmark result is very low and
error rate is very high. There seems to be some trouble in this mode.
- Here's what I did
- Changed from httpd to httpd.worker in /etc/sysconfig/httpd
- Disable PHP4
- LD_PRELOAD in /etc/init.d/httpd then start httpd
- The HTTPD start correctly and seems to work ok, but not all the time. Two
consecutive wget's on exactly the same URL yield different results
[root at compute-0-9 ~]# wget -v --header "Host: www.myhost.local" -O /dev/null
http://c0-3/cafe/siam/topic/F7800428/F7800428.html
--15:52:12-- http://c0-3/cafe/siam/topic/F7800428/F7800428.html
Resolving c0-3... 10.1.255.251
Connecting to c0-3|10.1.255.251|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 264373 (258K) [text/html]
Saving to: `/dev/null'
100%[====================================================================================================================>]
264,373 --.-K/s in 0.003s
15:52:12 (83.3 MB/s) - `/dev/null' saved [264373/264373]
[root at compute-0-9 ~]# wget -v --header "Host: www.myhost.local" -O /dev/null
http://c0-3/cafe/siam/topic/F7800428/F7800428.html
--15:52:15-- http://c0-3/cafe/siam/topic/F7800428/F7800428.html
Resolving c0-3... 10.1.255.251
Connecting to c0-3|10.1.255.251|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 0 [text/html]
Saving to: `/dev/null'
[
<=>
] 0 --.-K/s in 0s
15:52:15 (0.00 B/s) - `/dev/null' saved [0/0]
[root at compute-0-9 ~]#
(Note the returned content-length. Returned contents, when corrected, is
ok)
Here's the error log in booster log file
(First wget)
[2009-08-03 15:36:21] D [libglusterfsclient.c:1340:libgf_vmp_search_entry]
libglusterfsclient: VMP Entry found: /usr/home/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html: /usr/home/
[2009-08-03 15:36:21] D
[libglusterfsclient-dentry.c:381:libgf_client_path_lookup]
libglusterfsclient: resolved path(/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html) to
2626532/2650924
[2009-08-03 15:36:21] D [libglusterfsclient.c:1340:libgf_vmp_search_entry]
libglusterfsclient: VMP Entry found: /usr/home/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html: /usr/home/
[2009-08-03 15:36:21] D
[libglusterfsclient-dentry.c:381:libgf_client_path_lookup]
libglusterfsclient: resolved path(/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html) to
2626532/2650924
(second wget)
[2009-08-03 15:36:26] D [libglusterfsclient.c:1340:libgf_vmp_search_entry]
libglusterfsclient: VMP Entry found: /usr/home/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html: /usr/home/
[2009-08-03 15:36:26] D
[libglusterfsclient-dentry.c:381:libgf_client_path_lookup]
libglusterfsclient: resolved path(/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html) to
2626532/2650924
[2009-08-03 15:36:26] D [libglusterfsclient.c:1340:libgf_vmp_search_entry]
libglusterfsclient: VMP Entry found: /usr/home/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html: /usr/home/
[2009-08-03 15:36:26] D
[libglusterfsclient-dentry.c:381:libgf_client_path_lookup]
libglusterfsclient: resolved path(/www/
www.myhost.com/webdoc/cafe/siam/topic/F7800428/F7800428.html) to
2626532/2650924
In server logs, there are just connect-destroy-connection messages in the
log. Nothing particularily wrong to me.
No such failure occur with Apache+Prefork.
2009/8/2 Somsak Sriprayoonsakul <somsaks at gmail.com>
> After moving the embed configuration file out from glusterfs, now httpd
> boot up ok and the web seems to work now.
>
> However, comment posting is not working. It seems that the code that do the
> html modification is not working in GlusterFS context. I found that the code
> modify web page locally instead. So I think workaround for my case is to
> mount glusterfs with fuse at the same path as booster.
>
> I will give it another benchmark and see how it goes.
>
> 2009/8/2 Somsak Sriprayoonsakul <somsaks at gmail.com>
>
> Ok, I have a chance to run booster over 2.0.4
>>
>> Please find the attach file for my configuration
>>
>> I did configure boost and try simple ls over my Gluster file system.
>> Here's the output of ls -al
>>
>> [root at compute-0-3 ~]#
>> LD_PRELOAD=/usr/lib64/glusterfs/glusterfs-booster.so ls -l /gluster/www/
>> ls: /gluster/www/: Invalid argument
>> ls: /gluster/www/members.pantip.com: Invalid argument
>> ls: /gluster/www/cafe.pantip.com: Invalid argument
>> ls: /gluster/www/admin.pantip.com: Invalid argument
>> ls: /gluster/www/www.pantip.com: Invalid argument
>> ls: /gluster/www/passwd3.sql: Invalid argument
>> ls: /gluster/www/passwd.sql: Invalid argument
>> ls: closing directory /gluster/www/: File descriptor in bad state
>> total 129972
>> drwxr-xr-x 3 root root 8192 May 11 16:13 admin.pantip.com
>> drwxr-xr-x 5 root root 8192 May 18 11:11 cafe.pantip.com
>> drwxr-xr-x 3 root root 8192 May 11 18:48 members.pantip.com
>> -rw-r--r-- 1 root root 66654820 May 18 10:50 passwd3.sql
>> -rw-r--r-- 1 root root 66225769 May 18 10:33 passwd.sql
>> drwxr-xr-x 11 apache apache 8192 May 18 09:47 www.pantip.com
>> [root at compute-0-3 ~]#
>>
>> [root at compute-0-3 ~]#
>> LD_PRELOAD=/usr/lib64/glusterfs/glusterfs-booster.so cp /etc/issue /gluster/
>> [root at compute-0-3 ~]#
>> LD_PRELOAD=/usr/lib64/glusterfs/glusterfs-booster.so ls -l /gluster/issue
>> ls: /gluster/issue: Invalid argument
>> -rw-r--r-- 1 root root 47 Aug 2 14:57 /gluster/issue
>> [root at compute-0-3 ~]#
>> LD_PRELOAD=/usr/lib64/glusterfs/glusterfs-booster.so cat /gluster/issue
>> CentOS release 5.3 (Final)
>> Kernel \r on an \m
>>
>> [root at compute-0-3 ~]#
>>
>>
>> Despite all those errors, output seems to be fine
>>
>> And this is what inside my booster.log
>>
>> [2009-08-02 14:55:05] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:55:05] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:55:05] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:55:05] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:55:05] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:55:05] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:55:05] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:56:27] E [libglusterfsclient.c:4194:__glusterfs_stat]
>> libglusterfsclient: path lookup failed for (/hosts)
>> [2009-08-02 14:56:37] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>> [2009-08-02 14:57:00] E [libglusterfsclient.c:4194:__glusterfs_stat]
>> libglusterfsclient: path lookup failed for (/issue)
>> [2009-08-02 14:57:07] E [libglusterfsclient.c:2244:glusterfs_getxattr]
>> libglusterfsclient: invalid argument: value
>>
>> Then, I try to LD_PRELOAD apache (prefork). I change the target from
>> /gluster to /usr/home instead (the web application needs it). Then I tried
>> to strace the httpd process and found that httpd crash at the points where
>> httpd tried to read configuration file stored on Gluster volume (bad file
>> descriptor). I will try to move this configuration file some other places
>> and test again.
>>
>> 2009/7/31 Raghavendra G <raghavendra at gluster.com>
>>
>> Hi,
>>>
>>> On Thu, Jul 30, 2009 at 11:39 AM, Somsak
>>> Sriprayoonsakul<somsaks at gmail.com> wrote:
>>> > Thank you very much for you reply
>>> >
>>> > At the time we used 2.0.3, and yes we used stock Apache from CentOS. I
>>> will
>>> > try 2.0.4 very soon to see if it's work.
>>> >
>>> > For Booster, it seems not working correctly for me. Booster complains a
>>> lots
>>> > of error with plain 'ls' command (but giving the correct output). Also,
>>> with
>>>
>>> Can you mail those errors?
>>>
>>> > booster, Apache process refuse to start. I will try 2.0.4 to see if it
>>> > improves. If not, I will attach error log next time.
>>>
>>> logs are very much appreciated.
>>>
>>> >
>>> >
>>> > 2009/7/30 Raghavendra G <raghavendra at gluster.com>
>>> >>
>>> >> Hi Somsak,
>>> >>
>>> >> Sorry for the delayed reply. Below you've mentioned that you've
>>> problems
>>> >> with apache and booster. Going forward, Apache over booster will be
>>> the
>>> >> preferred approach. Can you tell us what version of glusterfs you are
>>> using?
>>> >> And as I can understand you are using apache 2.2, am I correct?
>>> >>
>>> >> regards,
>>> >> ----- Original Message -----
>>> >> From: "Liam Slusser" <lslusser at gmail.com>
>>> >> To: "Somsak Sriprayoonsakul" <somsaks at gmail.com>
>>> >> Cc: gluster-users at gluster.org
>>> >> Sent: Saturday, July 25, 2009 3:46:14 AM GMT +04:00 Abu Dhabi / Muscat
>>> >> Subject: Re: [Gluster-users] Gluster 2.0.3 + Apache on CentOS5
>>> performance
>>> >> issue
>>> >>
>>> >> I haven't tried an apples to apples comparison with Apache+mod_gluster
>>> vs
>>> >> Apache+fuse+gluster however i do run both setups. I load tested both
>>> >> setups
>>> >> so to verified it could handle 4x our normal daily load and left it at
>>> >> that.
>>> >> I didn't actually compare the two (although that might be cool to do
>>> >> someday).
>>> >> I really like the idea of Apache+mod_gluster as I don't have to deal
>>> with
>>> >> the whole fuse and mounting the filesystem. It always scares me
>>> having a
>>> >> public facing webserver with your whole backend fileshare mounted
>>> locally.
>>> >> Its very slick for serving content such as media files. We serve
>>> audio
>>> >> content to our CDN with a pair of Apache/mod_gluster servers - pushing
>>> >> 200-300mbit on average daily and everything works very well.
>>> >>
>>> >> We run an apache+fuse+gluster setup because we need to run some
>>> mod_perl
>>> >> before serving the actual content. However performance is still very
>>> >> good.
>>> >> We do around 50-100 requests (all jpeg images) per second off of a
>>> fuse
>>> >> mount and everything works great. We also have a java
>>> tomcat+fuse+gluster
>>> >> service which does image manipulation on the fly off of a gluster
>>> mount.
>>> >>
>>> >> We have two backend gluster servers using replication which serve all
>>> this
>>> >> content.
>>> >>
>>> >> If you would like more information on our setup id be happy to share
>>> >> offline. Just email me privately.
>>> >>
>>> >> thanks,
>>> >> liam
>>> >>
>>> >> On Fri, Jul 24, 2009 at 8:08 AM, Somsak Sriprayoonsakul
>>> >> <somsaks at gmail.com>wrote:
>>> >>
>>> >> > Oh thank you, thought noone will reply me :)
>>> >> >
>>> >> > Have you tried Apache + Fuse over GlusterFS? How is the performance?
>>> >> >
>>> >> > Also, anyone in this mailing-list have tried Apache with booster? I
>>> >> > tried
>>> >> > it but Apache refuse to start (just hang and freeze).
>>> >> >
>>> >> > 2009/7/23 Liam Slusser <lslusser at gmail.com>
>>> >> >
>>> >> >
>>> >> >> We use mod_gluster and Apache
>>> >> >> 2.2 with good results. We also ran into the same issue as you that
>>> we
>>> >> >> ran out of memory past 150 threads even on a 8gig machine. We got
>>> around
>>> >> >> this by compiling Apache using mpm-worker
>>> >> >> (threads) instead of prefork - it uses 1/4 as much ram with the
>>> same
>>> >> >> number
>>> >> >> of connections (150-200) and everything has been running smoothly.
>>> I
>>> >> >> cannot
>>> >> >> see any performance difference except it using way less memory.
>>> >> >> liam
>>> >> >>
>>> >> >>
>>> >> >> On Sun, Jul 12, 2009 at 5:11 AM, Somsak Sriprayoonsakul <
>>> >> >> somsaks at gmail.com> wrote:
>>> >> >>
>>> >> >>> Hello,
>>> >> >>>
>>> >> >>> We have been evaluating the choice for the new platform for a
>>> webboard
>>> >> >>> system.
>>> >> >>> The webboard is PHP scripts that generate/modify HTML page when
>>> user
>>> >> >>> posting/add comment to the page, resulting topic is actually
>>> stored as
>>> >> >>> a
>>> >> >>> HTML file with all related file (file attach to the topic, etc..
>>> >> >>> )stored in
>>> >> >>> its own directory for each topic. In general, the web site mostly
>>> >> >>> serve a
>>> >> >>> lot of small static files using Apache while using PHP to do other
>>> >> >>> dynamic
>>> >> >>> contents. This system has been working very well in the past, with
>>> the
>>> >> >>> increasing page view rate, it is very likely that we will need
>>> some
>>> >> >>> kind of
>>> >> >>> Cluster file system as backend very soon.
>>> >> >>>
>>> >> >>> We have set up a test system using Grinder as stress test tool.
>>> The
>>> >> >>> test
>>> >> >>> system is 11 machines of Intel Dual Core x86_64 CentOS5 with stock
>>> >> >>> Apache
>>> >> >>> (prefork, since the goal is to use this with PHP), linked together
>>> >> >>> with
>>> >> >>> Gigabit Ethernet. We try to compare the performance of either
>>> using
>>> >> >>> single
>>> >> >>> NFS server in sync mode against using 4 Gluster nodes (distribute
>>> of 2
>>> >> >>> replicated nodes) through Fuse. However, the transaction per
>>> second
>>> >> >>> (TPS)
>>> >> >>> result is not good.
>>> >> >>>
>>> >> >>> NFS (single server, sync mode)
>>> >> >>> - 100 thread of client - Peak TPS = 1716.67, Avg. TPS = 1066,
>>> mean
>>> >> >>> response time = 61.63 ms
>>> >> >>> - 200 threads - Peak TPS = 2790, Avg. TPS = 1716, mean rt = 87.33
>>> ms
>>> >> >>> - 400 threads - Peak TPS = 3810, Avg. TPS = 1800, mean rt = 165ms
>>> >> >>> - 600 threads - Peak TPS = 4506.67, Avg. TPS = 1676.67, mean rt =
>>> >> >>> 287.33ms
>>> >> >>>
>>> >> >>> 4 nodes Gluster (2 distribute of replicated 2 node)
>>> >> >>> - 100 thread - peak TPS = 1293.33, Avg. TPS = 430, mean rt =
>>> 207.33ms
>>> >> >>> - 200 threads - Peak TPS = 974.67, Avg. TPS = 245.33, mean rt =
>>> >> >>> 672.67ms
>>> >> >>> - 300 threads - Peak TPS = 861.33, Avg. TPS = 210, mean rt =
>>> 931.33
>>> >> >>> (no 400-600 threads since we run out of client machine, sorry).
>>> >> >>>
>>> >> >>> gfsd is configured to use 32 thread of iothread as brick.
>>> gfs-client
>>> >> >>> is
>>> >> >>> configured to use
>>> >> >>> io-cache->write-behind->readahead->distribute->replicate.
>>> >> >>> io-cache cache-size is 256MB. I used patched Fuse downloaded from
>>> >> >>> Gluster
>>> >> >>> web-site (build through DKMS).
>>> >> >>>
>>> >> >>> As the result yield, it seems that Gluster performance worse with
>>> >> >>> increasing no. of client. One observation is that the glusterfs
>>> >> >>> process on
>>> >> >>> client is taking about 100% of CPU during all the tests.
>>> glusterfsd is
>>> >> >>> utilizing only 70-80% of CPUs during the test time. Note that
>>> system
>>> >> >>> is Dual
>>> >> >>> core.
>>> >> >>>
>>> >> >>> I also tried using modglusterfs and not using fuse at all to serve
>>> all
>>> >> >>> the static files and conduct another test with Grinder. The result
>>> is
>>> >> >>> about
>>> >> >>> the same, 1000+ peak TPS with 2-400 avg. TPS. A problem arise in
>>> this
>>> >> >>> test
>>> >> >>> that each Apache prefork process used more about twice more memory
>>> and
>>> >> >>> we
>>> >> >>> need to lower number of httpd processes by about half.
>>> >> >>>
>>> >> >>> I tried disable EnableMMAP and it didn't help much. Adjusting
>>> >> >>> readahead,
>>> >> >>> write behind according to GlusterOptimization page didn't help
>>> much
>>> >> >>> either.
>>> >> >>>
>>> >> >>> My question is, there seems to be bottleneck in this setup, but
>>> how
>>> >> >>> can I
>>> >> >>> track this? Note that, I didn't do any other optimization other
>>> than
>>> >> >>> what
>>> >> >>> said above. Are there any best practice configuration for using
>>> Apache
>>> >> >>> to
>>> >> >>> serve a bunch of small static files like this around?
>>> >> >>>
>>> >> >>> Regards,
>>> >> >>>
>>> >> >>> Somsak
>>> >> >>>
>>> >> >>>
>>> >> >>>
>>> >> >>>
>>> >> >>> _______________________________________________
>>> >> >>> Gluster-users mailing list
>>> >> >>> Gluster-users at gluster.org
>>> >> >>> http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users
>>> >> >>>
>>> >> >>>
>>> >> >>
>>> >> >
>>> >>
>>> >> _______________________________________________
>>> >> Gluster-users mailing list
>>> >> Gluster-users at gluster.org
>>> >> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
>>> >
>>> >
>>>
>>>
>>> regards,
>>> --
>>> Raghavendra G
>>>
>>
>>
>
More information about the Gluster-users
mailing list