VPS Providers! What's Your Take On "write Back" & "write Through"
Oct 9, 2009
I recently worked on an issue involving a severe performance issue between "write back" and "write through" caching on the RAID/HD.
Long story short, we purchased 12 IBM x3550 M2s came with LSI SAS1068E/SR-BR10i (gimp redheaded stepchild of the MR series. No BBU, no onboard DIMM.) RAID controller and had very bad and inconsistent write throughput with it. Sometimes it writes out 300-400MB/s (dd test, I know... don't flame. I know dd is NOT a good test.), somethings as low as 30MB/s. The server is configured with 2.5" SATA 500GB on HW RAID-1.
From dmesg log it was default to "write through" on sda. I figured out via lsiutil, you can set the drives to "write back". Once we did that, the write performance is more consistent. NOTE, this enables write back on the SATA drives, NOT the controller itself.
I loop out lspci on all our VPS servers. Found out those with LSI (SAS 8344ELP) cards have sda set to "write through w/ FUA". Those are already all RAID-10s, and I have not heard a single complaint from any customer stating poor I/O performance.
I believe the 8344ELP do have a BBU, I can double-check with DC. The DC is on UPS as well. So that rules out the shortcomings of enabling write back caching.
I want to ask those of you using Xen (3.3 & 3.4). Do you guys have better I/O performance with "write back" or "write through" caching? I'm actually looking for real world results. Where you guys actually deployed production VPS server with clients on it.
config.php to not have write permissions for everyone? I am running cpanel 10x with whm/extras Here is full report when i try and load fantastico scripts. You must secure this program. Insecure permissions on config.php While installing CSLH you might of needed to change the permissions of config.php so that it is writable by the web server. config.php no longer needs to be written to so please chmod config.php to not have write permissions for everyone. you can do this by UNCHECKING the box that reads write permissions for the file:
I use a persistent and session cookie in order to help track return visitors, latent conversions and keyword tracking. My problem is that my hosting company, BlueHost, does not allow me to write the cookie data to my raw access log which would enable me to analyze this information.
Does anyone have any experience with using a script in order to get around this issue and write the cookie data to my log file....or as an alternative can people recommend a good hosting company which enables this functionality?
BlueHost has generally worked fine, so a work-around would be preferred, but if needed I may switch.
All is fine but the writes are very slow; I supose it's because write cache not enabled.
I have enable it and make some test and there is a lot of difference:
[url]
I'm worried about enabling it because if the server goes down I think it can cause disk corruption and loose data and maybe the OS. What do you think about that? Data is a priority!
Disks have this technology: Seagate-exclusive IRAW (Idle Read After Write) enhances data protection by verifying—during drive idle time—that data in thedrive buffer was properly written.
The servers are on a datacenter and have RAID 1 with Cheetah® 15K.5 SAS 3Gb/s 146.8-GB Hard Drive - ST3146855SS [url]
I'm also VERY worried about the coments about this card:
[url]
Is it true that the raid don't get rebuilded? If it is like this I don't know why I'm running raid.
I have learned some bits of regular expressions for simple scripting, writing a .htaccess file is, uh, syntaxically daunting.
THE CASE :
The URLs of my site used to be of the form [URL] ... . They are now of the form [URL]......
I am trying to perma-redirect (301) the old format (affiche_fiche.php) to the new format (fiche.php) using a .htaccess.
So far all I have achieved is a hatred of punctuation signs. What's the correct syntax to have a .htaccess that does the redirect ?
THE CONTEXT : The format change took place more than six months ago, but the Google Webmaster Tools still spits 450 problems a day with 404s on URLs using the old format. I had assumed that these would just fade away, but they don't. So I guess that 301'ing them is cleaner. Or would be, if I understood the syntax.
I just completed my latest site launch with BirdHosting.com and I have to say Mike knows how to operate this hosting company. I've been with ourinternet.us (spare me ever again---please), HostMySite, and GoDaddy. Godaddy was my latest host until I moved it over to Bird. Godaddy didn't give me much flexibility with their VPS windows hosting plans. I needed more then 10 gigs on a windows host.
A buddy of mine referred me to Mike from BirdHosting and Mike had me up and going before I even gave him a credit card. He really wanted to impress and meet all my needs. I now have a 50 gig / 1gig windows vps for a super great monthly price. I enjoy the personal attention I get when something goes wrong.
BirdHosting also helped me pick out two servers for where I work at too. Mike spent over 45 minutes on the phone with me answering questions I had. Godaddy would NEVER do that.
Anyways - I'm enjoying super FAST up time since going over to the birds nest. worth a look at if you are considering hosting companies.
I have a website built with asp .Net with MS SQL database. The website have some pages that uploads images and files in some folders.
I have tried GoDaddy shared Windows hosting. But they don't seem to allow file upload using a script.
Then I tried GoDaddy Virtual Dedicated. The tech guy told me MS SQL would be part of the hosting. Later I find he lied, there is no MS SQL as part of it.
I am trying to find an affordable hosting that has MS SQL and also allows file upload using script.
I just got 2 dedicateds, and while creating software raid 1, upon initial sync up I'm getting around 7 megabytes per second (6700 kb/s) in write speed I assume. This is a quad core, sata2 setup...
I'm using WAMP and IIS for the first time so bear with me if I overlooked something obvious.
I have a simple script that creates a few files and then attempts to move them to a subdirectory. I am getting a permission denied error when attempting to move the scripts in WAMP. I am uncertain how to set directory permissions in WAMP. I am also unsure how this will affect the script when I migrate it to an IIS server.
Warning: Unknown(): write failed: No space left on device (28) in Unknown on line 0
Warning: Unknown(): Failed to write session data (files). Please verify that the current setting of session.save_path is correct (/tmp) in Unknown on line 0 I am getting this error on my site. I have googled this error and it is telling me it relates to a /temp/ folder of some form. I am currently on a hosting plan (not a dedicated server). Is there ANY way I can access to fix this problem either from my control panel or by code?
I have a concern during the installation of PrestaShop. I created a subdomain through Plesk, put my files on my server via FTP. During installation of PrestaShop, I have a problem:
Recursive write permissions for the Apache user on the ~ / config /
My user is "NOWIS" and my group is "psacln" ....
I think the user "NOWIS" has no rights as "Apache" ....
I just discovered there is seperate error log for php-fpm and since only 1 site is running fully nginx, the log is full with this error:
"NOTICE: PHP message: PHP Warning: Unknown: Failed to write session data (files). Please verify that the current setting of session.save_path is correct (/var/lib/php/session) in Unknown on line 0"
I just upgraded our Plesk 10.4.4 to 12.0.18 (on Ubuntu 10.04, about to be upgraded to 12.04). Now I get the dreaded 'queue file write error' in postfix on message delivery.
It tries to open /usr/lib/plesk-9.0/postfix-queue (which is not there) in this block in master.cf:
Code:
plesk_virtual unix - n n - - pipe flags=DORhu user=popuser:popuser argv=/usr/lib/plesk-9.0/postfix-local -f ${sender} -d ${recipient} -p /var/qmail/mailnames 127.0.0.1:10025 inet n n n - - spawn user=mhandlers-user argv=/usr/lib/plesk-9.0/postfix-queue 127.0.0.1 10027 before-queue 127.0.0.1:10026 inet n - - - - smtpd -o smtpd_client_restrictions= -o smtpd_helo_restrictions= -o smtpd_sender_restrictions= -o smtpd_recipient_restrictions=permit_mynetworks,reject -o smtpd_data_restrictions= -o receive_override_options=no_unknown_recipient_checks 127.0.0.1:10027 inet n n n - - spawn user=mhandlers-user argv=/usr/lib/plesk-9.0/postfix-queue 127.0.0.1 10026 before-remote plesk_saslauthd unix y y y - 1 plesk_saslauthd status=5 listen=6 dbpath=/plesk/passwd.db
And on top of that, If disable all the milters/proxy and just try to deliver to plesk_virtual directly, the e-mail just disappears. Postfix thinks it was delivered, but it is not.
I also tried piping some text through /usr/lib/plesk-9.0/postfix-local with the same parameters postfix gives it, but also then nothing. No error, no exit code, no message in maildir.
There are a bunch of fixes around for the queue file write error, but nothing works because the nature of my problem is different. There is a KB article around for Plesk 9, which can hardly be useful.
I tried restoring postfix-queue from the backups, but if I try to use that, the mail.log eventually says:
Code:
Jan 30 01:20:00 ytechosting /usr/lib/plesk-9.0/psa-pc-remote[6621]: Message aborted.
I also used /usr/local/psa/admin/sbin/mchk to no avail: the only change it makes to my master.cf, is adding an empty line. And, I suspect it uninstalled my psa-courier, because it was gone at some point, and mchk was the only thing I did after knowing courier was still there.
How can the postfix config refer to that postfix-queue when it's not there? Is this even a valid master.cf file for Plesk 12?
we've been testing CentOS 5.3 on Intel DG35EC board (G35+ICH8+82566 Gb NIC), and found that the write speed out of 7200rpm SATA-II drive, connected to on-board ICH8 controller, is consistently under 10MB/sec which is quite horrible!
the same set of hardware can get 100MB+/sec transfer rate with Debian 5.0 and FreeBSD 7.1, just not the CentOS5.3! it doesn't matter whether AHCI mode is selected in BIOS or not, and of course BIOS has been updated with latest version.
Before i start i would just like to mention i had this problem about 8-10 months ago but fought this would be useful to share.
I bought the hostgator swamp shared hosting package for a website me and my friend were going to make.
Long story short my friend didnt have to much time in his hands so pulled out so a was left with hosting that i didnt need, so about 30-35 days in i cancelled the account, Hostgator claim that if you cancel your account with them no more than 45 days being with them the will issue a full refund.
I waited the full 3-5 business days they say it takes to issue the refund, I never recieved that refund, i gave them another week still no sign of the money.
After numerous chats with live operaters and opening tickets, i still hadn't got anywhere, live operaters where giving me coupons to make up for the refund ( but thats still not good enough ) and the operaters answering the tickets were saying they had issued the refund.
I just decided to forget about it.
The service was good though in that period of time they, such as changed my main domain name for me for free because i was a new customer.
