Diagnostic Help

[Gauravk]

Hi All,

I have recently migrated the community to IPB from PHPBB and apparently facing triple the load time.

On same server (2GB Ram, 8 core Xeon) in phpbb it was taking around 2-2.5 secs and now with IPB its taking around 6-8 secs. I am running classifieds, company directory and IP content. I tried switching all off and still load time reduced by 10% to 5.5 - 7 secs nothing really noticeable difference.

I installed lazy loader install but no luck, its still running.

Appreciate if someone with more knowledge can help me finding the real pain point of this slow load up.

Domain name: carnity dot com

Thanks in advance.

[PeterUK]

You need to answer a few more questions for the best help here.

What's your technology stack (ie. Apache, or nginx + php-fpm, for example)? 2GB RAM also seems awfully low for an 8 core Xeon, however that's not to say it should have any problems running a decent sized IPB installation.

How large is your board, how many posts, topics and concurrent users in general?

[Dmacleo]

this a vps?

wondering if its drive speed related

[Gauravk]

Its a Centos with Apache and PHP 5.5. on VPS.

Board has about 12k members, 25k post.

MySQL was optimized for Innodb for PHPBB in past, and recently optimized for MyIsam to serve IPB but still no much improvement felt.

[Gauravk]

Gentle reminder

[Grumpy]

I don't really know what you are using to get the measurement you posted above, but seems to load faster for me, even if I include every content (except ads).

Regardless, it's not fast, that is clear. Whether or not IPB requires more resource to run than PHPBB, is a question I don't know the answer to. So, it's possible that your slower response rate is simply a result of running something heavier.

I recommend you to follow general optimization techniques first. There are ton of those threads here. Which generally involves:

• Installing php opcache (like zend opcache)
• Installing user cache (like APCu)
• Installing sphinx (for search)
• Swapping from myisam to innodb on table locking forums. (though I see you swapped to myisam...)
• Swapping session table to memory engine.
• Swapping out apache for nginx & php-fpm. Or if you need a quicker solution, nginx or varnish reverse proxy to apache.

And general advice/note for VPS is that:

• VPS is unreliable.
• Try using SSD backed VPS.

If it's not possible for you to complete the optimization yourself, there are people here available for hire as well.

[Grumpy]

Whelp... realized I didn't answer the actual question.

Diagnosing.

Slowness or time to process is a really a question of what resource you can afford. Commonly deriving down to cpu, memory, disk & network. You identify your bottleneck (the one you are most lacking) and attempt to resolve that.

You can try running commands like top in ssh. That will give you a good look at the first 3: cpu, memory & disk.

If the load value is too high, it's safe to say you are lacking some sort of resource.

If the user + system is too high, you need better/more processing power OR reduce need to process.

If the memory used minus cache & buffer is too high, you need more memory. Additionally, if you are hitting swap, you need more memory OR reduce the need to consume so much memory.

If the io wait (%wa / alternatively %hi + %si [hardware & software interrupt]) is too high, then it's likely that your disk is your problem. Install better/more disks in raid OR reduce the need to access disk.

Optimizing is basically reducing the need of a certain resource(s).

If your processing is actually fast, but slow to the user, then you have a network problem.

You can reduce a resource usage by suggestions like the last post. It's also possible that a bad script is requiring excessive resources. Like hooks, plugins, some skins, etc. You can identify them by disabling them one by one. If you have a ton of them, you can isolate problematic ones faster by methods like the binary search.

[Gauravk]

Thanks a lot Grumpy for lot of inputs and great ideas. You rock.

I will explain, what exactly I have in my mind. Few years back when I was in similar situation and as general public like to shoot the server guys first by saying oh your server is slow. By listening to few people we switch our server and realized that PHPBB forum still loading in 8 seconds time. Later on some wise guy like you advise to see MySQL my.cnf and we realize that main bottleneck was MySQL as nothing was specified in that file. It means that no matter we were on 2GB ram, but MySQL using default 128MB only and that is causing the first 6 seconds to think and process the query. That wise guy helped us optimizing MySQL and pages started loading in 2 seconds from any location. Sheer genius.

Learning from above experience, my intention is to actually identify the root cause of this slight delay and try to optimize it as much as possible. We don't mind switching servers, adding rams, going for dedicated but only if we have to.

Please find the latest stuff below and see what can be the possible bottleneck in our scenario. Really appreciate your help. Thanks in advance.

SSH TOP output

top - 12:32:53 up 113 days, 14:17,  1 user,  load average: 0.42, 1.50, 1.68
Tasks:  90 total,   2 running,  86 sleeping,   0 stopped,   2 zombie
Cpu(s):  3.5%us,  0.6%sy,  0.0%ni, 95.9%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   2097152k total,  1432488k used,   664664k free,        0k buffers
Swap:        0k total,        0k used,        0k free,        0k cached

Current my.cnf

Small note on below config, that I have been optimizing this by my very limited knowledge of MySQL since 3 days by listening to advice from mysqltuner.pl on daily basis. Its getting better but still not quite fast as I expect it load under 2 seconds.

Secondly innodb tables are the one from phpbb board, which we are not using anymore, and will remove those tables in future completely. Hope that bulk of 332MB is not causing any delay as those urls are redirected to IPB version now and that's why I have just allocated 32 megs for innodb.

Thirdly, we dont mind switching to innodb (In end), but we like to try myisam as much as possible for full text search functionality, as I really like the idea of showing related topics under every topic view page.

[mysqld]


#default_storage_engine=InnoDB
default-storage-engine=MyISAM


max_connections=50


key_buffer_size = 64M
max_allowed_packet=268435456


query_cache_size = 256M
query_cache_limit = 4M
tmp_table_size = 256M
thread_cache_size = 256
max_heap_table_size = 256M


myisam_sort_buffer_size=128M


table_cache = 256M
table_definition_cache = 4096


slow_query_log
slow_query_log_file=mysql-slow.log
long_query_time=1
#log-queries-not-using-indexes


join_buffer_size = 2M


low_priority_updates=1
concurrent_insert=ALWAYS


thread_concurrency = 8


#### Per connection configuration ####
sort_buffer_size = 2M
read_buffer_size = 2M
read_rnd_buffer_size = 4M


##### INNODB Specific Options ######
innodb_buffer_pool_size=32M
innodb_log_file_size = 50M
innodb_flush_log_at_trx_commit=0
innodb_log_buffer_size=8M
innodb_flush_method=O_DSYNC
#innodb_old_blocks_time=1000
innodb_open_files=2000


innodb_file_per_table=1
open_files_limit=20062

Current mysqltuner.pl report

 >>  MySQLTuner 1.2.0 - Major Hayden <major@mhtx.net>
 >>  Bug reports, feature requests, and downloads at http://mysqltuner.com/
 >>  Run with '--help' for additional options and output filtering


-------- General Statistics --------------------------------------------------
[--] Skipped version check for MySQLTuner script
[OK] Currently running supported MySQL version 5.5.37-cll
[OK] Operating on 32-bit architecture with less than 2GB RAM


-------- Storage Engine Statistics -------------------------------------------
[--] Status: +Archive -BDB -Federated +InnoDB -ISAM -NDBCluster
[--] Data in MyISAM tables: 119M (Tables: 895)
[--] Data in InnoDB tables: 332M (Tables: 3793)
[--] Data in PERFORMANCE_SCHEMA tables: 0B (Tables: 17)
[--] Data in MEMORY tables: 0B (Tables: 2)
[!!] Total fragmented tables: 237


-------- Security Recommendations  -------------------------------------------
[OK] All database users have passwords assigned


-------- Performance Metrics -------------------------------------------------
[--] Up for: 12h 28m 58s (477K q [10.632 qps], 25K conn, TX: 20B, RX: 84M)
[--] Reads / Writes: 54% / 46%
[--] Total buffers: 624.0M global + 10.2M per thread (50 max threads)
[OK] Maximum possible memory usage: 1.1G (55% of installed RAM)
[OK] Slow queries: 0% (212/477K)
[OK] Highest usage of available connections: 30% (15/50)
[OK] Key buffer size / total MyISAM indexes: 64.0M/39.9M
[OK] Key buffer hit rate: 99.7% (1M cached / 6K reads)
[OK] Query cache efficiency: 70.9% (196K cached / 277K selects)
[OK] Query cache prunes per day: 0
[OK] Sorts requiring temporary tables: 0% (0 temp sorts / 6K sorts)
[!!] Temporary tables created on disk: 28% (17K on disk / 62K total)
[OK] Thread cache hit rate: 99% (15 created / 25K connections)
[!!] Table cache hit rate: 1% (9K open / 718K opened)
[OK] Open file limit used: 13% (2K/20K)
[OK] Table locks acquired immediately: 99% (214K immediate / 214K locks)
[!!] InnoDB data size / buffer pool: 332.5M/32.0M


-------- Recommendations -----------------------------------------------------
General recommendations:
    Run OPTIMIZE TABLE to defragment tables for better performance
    MySQL started within last 24 hours - recommendations may be inaccurate
    Temporary table size is already large - reduce result set size
    Reduce your SELECT DISTINCT queries without LIMIT clauses
    Increase table_cache gradually to avoid file descriptor limits
Variables to adjust:
    table_cache (> 524288)
    innodb_buffer_pool_size (>= 332M)

[Grumpy]

Your top output actually seems to have missed the more important timing. As i see 15min avg being the highest of the 3 load averages. So, something heavy was there, and no longer is there. It's also more helpful to post the entire top, not just the first part.

If IPB is the only thing you are running, don't increase your table cache any further. Actually decrease down to 32M. Mysqltuner always gets this wrong with IPB because of the types of temp. tables IPB makes. It'll just store junk in there and waste more time shifting through the junk repeatedly. Even with decreased amount, your prune rate will likely not change, in other words, your table cache hit rate will always suck. But by decreasing it down to 1/8th of your prev size, you'll effectively reduce ~3 steps of wasting time in cache (assuming 2 branch b-tree, which it is not...).

Note. If you do opt to use MEMORY engine for sessions table, check the size of it first. It must fit inside luxuriously under the table heap size.

IF at the moment of the top capture, your forum still took a long time to load: I can only assume your vps is faeces. You are not bottlenecked on anything and you are hardly using your share of resources. Yet it takes long time. That can only mean your underlying vps hardware is bottlenecked at someone/something other than you. That could be the hardware or the network. As a vps container, it's not possible for you to see the true hardware statistics.

You can still take above advices in improving server performance. It will make a significant difference. But, won't change the fact that your vps is still faeces.

max_allowed_packet=268435456

Max allowed packet is not an optimization but an anti-crazy limit. 256MB query sounds pretty crazy (considering we're not in big data realm). Imagine a query/result so long that the single query/result is longer than your average novel. How big is 256MB? This skyrim walkthrough is 999KB. 256MB is 256 times that length.

p.s. faeces is this forum's auto-censor to a word that is really not that much worse. lol

[Gauravk]

Thanks Grumpy,

For detailed response, I have edited table cache to 32M and max allowed packet to 2M. As you guessed for excessive table cache right reason was: mysqltuner.pl. However for max packet, in past I need to upload bigger sql dump and I think I increased at that time.

Please find the 6 screen shots of Top output, 3 under slight load and 3 normal. btw, I have edited the server name to carnity01 to _______.

I appreciate you taking so much time in helping in diagnosing the real bottleneck, based on which I can pin point to server provider with adequate details if there VPS is really a faeces. lol. (after long time I heard this word in english).

We also need to remember and being fair that current page size is 1.3MB tested by pingdom, gtmetrics and yslow as compared to previous phpbb just under 500kb.

May be you can also see if I really have to optimize the page as per yslow advice and join different CSS and JS, reduce dns lookup etc...? As when I started using browser cache transfer reduced to 400KB from 1.3 MB. Wish IPB htaccess already have such lines for caching images and other static scripts.

[Machsterdaemon]

Use:
table_open_cache

instead of

table_cache (deprecated variable)

The default value for your version of MySQL is 400. I'm not sure if you can safely start with high value like 5000, or you should start with some lower value and increase it slowly in 2 or 3 steps and start with this:

table_open_cache=2000

Remove:
thread_concurrency

It's probably not needed for your OS.

Increase:
innodb_buffer_pool_size=32M
> 512M

Decrease:

tmp_table_size = 256M
max_heap_table_size = 256M

Try:

tmp_table_size = 64M # or try 32M
max_heap_table_size = 64M # or try 32M

Decrease:

query_cache_size = 256M
query_cache_limit = 4M

Try:

query_cache_size = 32M # or 16M
query_cache_limit = 1M

Or just remove that variable from the option file to disable it.

[Gauravk]

Thanks Machsterdaemon,

Appreciate your help and advice's.

We are not using innodb tables any more, its just the data sitting there as a backup of our old community. Should we still increase innodb value...? Or remove the db completely to gain some performance...?

[Machsterdaemon]

Thanks Machsterdaemon,

Appreciate your help and advice's.

We are not using innodb tables any more, its just the data sitting there as a backup of our old community. Should we still increase innodb value...? Or remove the db completely to gain some performance...?

You don't need to increase the innodb bufffer pool size if you have no use for it, and I don't thinks it's necessary to remove the innodb database. Removing it would just help with mysql server manintenance and server backup I think.

If you follow any mysql tuning related advice in this topic now, I think you should run mysqltuner again within 6-12 hours just to make sure that changed configuration is not making the server perform worse.

When 24-48 hours have passed, you should run mysqltuner again and post the output in this topic.

You should perhaps use another tool like mysqlreport and read the guide.

Update:

You can use:

table_open_cache=1000 # or higher

[Grumpy]

Are you in signapore? Are your majority users also in singapore? I'm confused now. Because you didn't mention anything about this yet.

Your server is in singapore. Hosted by godaddy, who is very crappy (even outside of sg). The network connection is so terrible I can't get beyond dial-up speeds from various locations around the world. I'm also seeing packet loss across the board, but possible that might be firewall caused.

There's no point of diagnosing anymore. Your server is faeces.

[Dmacleo]

notice both your nameservers are the same ip.

possible your nameservers are getting enough traffic to slow response?

yeah ping times are horrible and I see loss too.

[Gauravk]

Grumpy, you got a good sense of humor mate. Yes server is in Singapore as it connects to rock solid and stable link between these two countries. I am using this same location server since 4 years and my PHPBB use to load in Dubai under 2 seconds on same server for 500 kb homepage without CDN. Looks like (I assume) your ISP have some lousy hops to Singapore link.

IPB homepage is about 1.3 MB and that's why its causing load around 5-6 seconds now from Dubai. My ping test showing 101 -109 ms for all 4 instance with 0 loss.

Dmacleo, you made a very good observation and I should work on it to assign two IP for different NS records. Thanks.

Machsterdaemon: I have accepted all you recommendation and even made table_cache at 2k and then bumped to 5k after 6 hours. Please find below latest mysqltuner report. Looks like we need to tweak some more settings this time.

Btw, yesterday I tried installing APC on my server and added one line code to activate APC in IPB config_global.php, but site refused to load completely and just showed blank white screen. Eventually I have to uninstall APC in end, any idea if any more additional settings or changes required in IPB installation....?

Really appreciate everyone attention and help on this subject. Thanks a lot guys.

-------- Storage Engine Statistics -------------------------------------------
[--] Status: +Archive -BDB -Federated +InnoDB -ISAM -NDBCluster
[--] Data in MyISAM tables: 121M (Tables: 895)
[--] Data in InnoDB tables: 330M (Tables: 3779)
[--] Data in PERFORMANCE_SCHEMA tables: 0B (Tables: 17)
[--] Data in MEMORY tables: 0B (Tables: 2)
[!!] Total fragmented tables: 245


-------- Security Recommendations  -------------------------------------------
[OK] All database users have passwords assigned


-------- Performance Metrics -------------------------------------------------
[--] Up for: 20h 2m 21s (874K q [12.116 qps], 47K conn, TX: 37B, RX: 161M)
[--] Reads / Writes: 54% / 46%
[--] Total buffers: 208.0M global + 10.2M per thread (50 max threads)
[OK] Maximum possible memory usage: 717.4M (35% of installed RAM)
[OK] Slow queries: 0% (369/874K)
[OK] Highest usage of available connections: 34% (17/50)
[OK] Key buffer size / total MyISAM indexes: 64.0M/40.8M
[OK] Key buffer hit rate: 99.8% (4M cached / 8K reads)
[OK] Query cache efficiency: 76.4% (432K cached / 565K selects)
[!!] Query cache prunes per day: 18401
[OK] Sorts requiring temporary tables: 0% (0 temp sorts / 9K sorts)
[!!] Temporary tables created on disk: 26% (19K on disk / 74K total)
[OK] Thread cache hit rate: 99% (17 created / 47K connections)
[!!] Table cache hit rate: 0% (4K open / 1M opened)
[OK] Open file limit used: 9% (1K/20K)
[OK] Table locks acquired immediately: 99% (351K immediate / 352K locks)
[!!] InnoDB data size / buffer pool: 331.0M/32.0M


-------- Recommendations -----------------------------------------------------
General recommendations:
    Run OPTIMIZE TABLE to defragment tables for better performance
    MySQL started within last 24 hours - recommendations may be inaccurate
    When making adjustments, make tmp_table_size/max_heap_table_size equal
    Reduce your SELECT DISTINCT queries without LIMIT clauses
    Increase table_cache gradually to avoid file descriptor limits
Variables to adjust:
    query_cache_size (> 32M)
    tmp_table_size (> 64M)
    max_heap_table_size (> 64M)
    table_cache (> 5000)
    innodb_buffer_pool_size (>= 330M)

[Aiwa]

You can't always trust the metrics places... Trust what you SEE in terms of speed and find what you can adjust on your server for performance. If your VPS node isn't the bottleneck your host is..

The VPS I'm on has 1G RAM, SSD, 4 CPU cores, clocks in at 2.0MB for my board index and pages exec in under 0.2 seconds, Exec. Time: 0.1794 on an IP.Content front page with 50+ queries... FORUM index page, Exec. Time: 0.0659, 8 queries. But it's with a VPS host that is known for great performance. GoDaddy, not so much.

http://www.tacticaladvantage.com/

[Machsterdaemon]

Machsterdaemon: I have accepted all you recommendation and even made table_cache at 2k and then bumped to 5k after 6 hours. Please find below latest mysqltuner report. Looks like we need to tweak some more settings this time.

table_open_cache
I did first mention a value of 5000 for table_open_cache as I overlooked that your other database was not in use. But I updated my later reply with a value of 1000 or higher.

table_open_cache=1000

I'm not sure if you should be using a value of 5000 if you don't have that many tables.

I mentioned the mysqlreport tool that I prefer to use more than mysqltuner when making changes. In the table report you get some information of your cache utilization that might be useful.

You not only get the table cache usage, but also the rate of opening tables. It can be of help in improving the table cache use.

wait_timeout

It seems that your mysql server is using the default value for the wait_timeout variable which is 28800 seconds. This is too high and it can influence your mysql server performance.

It's generally recommended to set this to a very low value to let the server close idle connections, like this (30 seconds value is just an example):

wait_timeout=30

Which means that idle connections for more than 30 seconds will be closed, instead of the default value of 28800 seconds. Many have this set to a value of 15 seconds.

Be aware of timeout errors when you set this this to a low value. You can increase the 10 seconds default value for the connect_timeout variable if you get timeout errors, like this:

connect_timeout=15

[Grumpy]

I have multiples of of servers around the world, all under 1gbps connection. Locations being: Atlanta; Richmond, Virginia; Chicago; San Jose; Toronto; 2 separate dc in Amsterdam; Frankfurt, Germany; etc. I'm not saying "my connection to your site is bad", I'm saying your server connection is bad. I checked the speed to a biggest static file on your server I could find when I tested yesterday. Each one of them were in ~50kbps range. None of them even went into 70kbps+... again, these are from a gigabit servers where I have the ability to burst them any time to a full gbps.

Exception is if your users are in singapore/close to, then it probably is fair at least, and I have an inappropriate sample. But you aren't saying you need singapore at this point. You even mention you are in Dubai right now. I don't really understand why you have one there. Asia has terrible network peering and pricing.

Homepage is indeed 1.3MB from zero cache. But after caching, browsing only takes ~100KB. That matters more.

The time to first byte (TTFB) on any given page is ~2s right now. And then I need another 2s to load 100KB (at ~50kb/s). That makes the total of 4s to load your site's page. If the connection was good, the TTFB would've been decreased, and the 2s to download would be reduced to <0.1s. The site would have loaded in <2s. Even downloading 1.3MB shouldn't take me seconds.

If your PHPBB was smaller in size and resulted in faster download time, that just explains the status of network issues even more. It says that the network is so crappy, it can't even reasonably send 700KB more.

Look, I don't consider 2s load of your past performance as good either. I expect a site to be done processing in <200ms and I expect it to be delivered in <200ms as well. So your entire site should load in <500ms. With heavy optimization (and some money), you can get IPB as low as <50ms to process and <100ms to download. Making the entire site load on your page in <0.15s. 2s goal isn't something you should be striving for, it's under <0.5s. You will never achieve that with your server in singapore because the download time alone passes the goal.

[Gauravk]

Thanks Machsterdaemon,

I turned back table cache to 1024 last night, awaiting for more 6-7 hours to complete 24 hr report. I also added wait_timeout=15 and so far server looks slightly more happier, no unusual stuff noted.

I tried installing mysqlreport couple of times and it kept on failing in root and in under cd /usr/local/sbin. wget is getting it fine chmod work as well but when I run ./mysqlreport it produce below errors

./mysqlreport
install_driver(mysql) failed: Can't load '/usr/local/lib/perl5/site_perl/5.8.8/i686-linux/auto/DBD/mysql/mysql.so' for module DBD::mysql: libmysqlclient.so.16: cannot open shared object file: No such file or directory at /usr/local/lib/perl5/5.8.8/i686-linux/XSLoader.pm line 71.
 at /usr/local/lib/perl5/site_perl/5.8.8/i686-linux/DBD/mysql.pm line 15
Compilation failed in require at (eval 9) line 3.
Perhaps a required shared library or dll isn't installed where expected
 at ./mysqlreport line 249

I also tried installing some more stuff required from here: but it still produce same error. Please advice what am I missing...?

Grumpy: My Car Community (http://carnity.com) is based out of Dubai ( I am also living in Dubai) and have about 90% users from middle eastern countries around Dubai, UAE. The reason I chose the Godaddy Asia pacific datacenter as that was loading blazing fast pages in Dubai. Earlier I was using US datacenter and initial lag was quite noticeable of upto 2-3 seconds (TTFB). Since we switch to Singapore datacenter speed improved noticeably.

I fully understand your point on server network speed now (thanks for explaining in lay man language), Is there is any way I can measure that and record and show it to Godaddy and ask them to improvise this....? You seem to be quite experienced in handling many server with different companies, do you mind making few suggestions for VPS and dedicated servers, that I can simultaneously pursue and evaluate if this server doesn't work out good.

[Machsterdaemon]

I'm not sure about the mysqlreport issue, but it's likely you don't have the perl module Term::ReadKey installed, but I'm not sure.

Have you tried to run the script with the password like this:

./mysqlreport --user root --password

and then entering the password when prompted.

[Grumpy]

If you want to provide proof of slow connection to your ISP, make a dummy file on your server. 10 ~ 100MB will do. Just use dd command with input as /dev/zero if you have nothing to get it from. like: dd if=/dev/zero of=foobar count=1 bs=100M

Then place the foobar file a public folder and have it downloaded from other servers using wget. Multiple times and multiple locations.

--------------------------

Unless getting a server in local area, the only two options are really NA & EU for getting a server. Because the others have terrible connection to the rest of the world and also much more expensive. Your best chance is probably in Romania. Though, I wouldn't stay off close countries like Germany. Not all DC are the same even if they're close by. Whether or not they optimize their traffic to ME would make a greater factor. So, it's important to test first. You can just get one in Dubai, it'll likely serve you and your large user base the best, but it'll be more pricy & further outside users like US will have a difficult time.

If you are shopping for a new server, you should check ping:

http://cloudmonitor.ca.com/en/ping.php

You'll need to ask them for a public IP first. This will only tell you of how fast it'll respond. Not throughput. Also note, the above ping tool will only work if they don't have it blocked via firewall. They usually don't firewall test IPs since thats dumb...

So, you also need to ask them for a test file. They will give you a link to a 100MB file or similar. You can try downloading that and check the network speed yourself.

If they cannot comply with either, you probably shouldn't go with them.

----------------------

As other hosting services are in competition with IPS, recommending here is not allowed. I can, however, recommend you to do your shopping in webhostingtalk.com. It's the best place for it anyway.

[Gauravk]

Mach, yes I did same again and got below error

./mysqlreport --user root --password
Password for database user root:
install_driver(mysql) failed: Can't load '/usr/local/lib/perl5/site_perl/5.8.8/i                                       686-linux/auto/DBD/mysql/mysql.so' for module DBD::mysql: libmysqlclient.so.16:                                        cannot open shared object file: No such file or directory at /usr/local/lib/perl                                       5/5.8.8/i686-linux/XSLoader.pm line 71, <STDIN> line 1.
 at /usr/local/lib/perl5/site_perl/5.8.8/i686-linux/DBD/mysql.pm line 15
Compilation failed in require at (eval 9) line 3, <STDIN> line 1.
Perhaps a required shared library or dll isn't installed where expected
 at ./mysqlreport line 249

Grumpy, thanks for the heads up. May I ask another help, while diagnosing this issue I was quiet focussed on TTFB today and realized that my domain redirects to non www version, this made me recall that sometime in past I have set this redirection in getting some edge over SEO stuff. I went to cpanel and check all possible redirection and I couldnt find one, then I check in htaccess and there is no reference to www. Then I want to create one to bring back to standard shape and made a cpanel redirection of redirecting non www to www version and ended up with redirection loop error in browser.

Can you think or advice how in the past (possibly) I have redirected my www to non www other than cpanel and htaccess? As I really want to get rid of that setting first as my few other dormant domain on same server showing 0.5 secs for TTFB.

[Dmacleo]

conf_global.php have address as non-www ?

[Gauravk]

Yes, that has w/o www because domain auto redirect has been active from quite sometime like a year back, when it was on different cms system.

I tried changing from conf_global just now, it loaded the site at www version but when I set the 301 redirection from cpanel for same, again went to loop :-(