|Frequently Asked Questions and Answers
Company | Email and web hosting accounts on shared servers |
Dedicated servers |
Virtual Private Servers | cPanel & WHM |
SPAM and SPF | Migration |
Troubleshooting for dedicated servers |
1. Which data centre are the servers in?
We have 12 racks in Qala. We have annual contracts with Qala for the racks.
2. Are you a reseller for the hosting accounts? i.e. you buy reseller accounts, and
resell hosting accounts
No, we get rack space direct from the data centers (Qala), and own our own servers.
We purchase software licenses direct from cPanel (for shared hosting) and Virtuozzo
(for VPS servers). We are an authorized cPanel Partner NOC, and Parallels Gold Partner.
3. Can customers view the servers or data centre?
Yes, please email firstname.lastname@example.org or give us a call to arrange for a data centre visit.
4. Is Xssist Group registered with ACRA?
Yes, Xssist Group Pte Ltd is incorporated in Singapore. The registration number is 200409501D.
The paid up capital is $20K.
Email and web hosting accounts on shared servers
1. What is the specifications of the server used?
Our current shared hosting servers are Quad Opterons, with 4GB ECC RAM, 146GB SCSI
harddisks (hardware RAID 1).
2. Do you have cheaper hosting on cheaper servers, eg. with 1 CPU and with SATA
or IDE harddisks?
Currently, we use only 4-CPU servers and SCSI harddisks for performance and reliability.
Our servers cost SGD10,000 to SGD20,000 each instead of the SGD500 to SGD3,500 servers used
by other companies. Our servers have 3 x redundant power supplies each, hardware RAID 1
(mirror) for the harddisks and hotswappable (SCA2) SCSI harddisks.
3. Why do you use RAID 1 instead of RAID 5?
RAID 1 has better performance compared to RAID 5, in particular, the write performance
is much better. As web and email servers does a lot of writes, to log files, email spools etc,
RAID 5 is not suitable for the server.
In terms of redundancy, RAID 5 has 1 parity drive in an array (of possibly 7 drives).
With a failure of 1 drive, the array performance is badly degraded. This may lead to much
higher IO waits, and the server will be unresponsive. Upon replacement of the failed drive
either manually or through a hot spare, the rebuild starts. It typically takes up to 1 day
or more to rebuild an array, as the failed array needs to work at the same time, in a
For RAID 1, only 2 drives are used. Read and write performance is good.
Although hardware RAID 1 is more expensive per GB, we believe in providing as much redundancy
as possible in our hardware, to minimize hardware downtime.
4. Why not RAID 0 for performance?
RAID 0 stripes data across several disks. This results in very good read and write
performance. However, failure of any 1 drive will lead to data loss.
Instead of using RAID 0, we can setup a second RAID 1 drive, and split half of the
accounts over to the second RAID 1 drive. This will distribute the disk IO
over the 2 RAIDs, with good data redundancy.
5. How about RAID 0 + 1 or RAID 10?
This gives good results but at very high prices as we use SCSI drives. Instead of doing this,
we simply setup a second server. This allows better performance. There is diminishing
returns in improving only 1 aspect of a server. The other parts of the server become the
bottleneck. It is usually better value for us to setup a second server, than to have
a RAID 10 setup, in a webhosting server. Your mileage may vary, since other factors
like availability of system administrator, and costs of licenses come into play.
6. What is the control panel used?
We use cPanel. We are experienced in cPanel and we are an authorized cPanel Partner NOC.
7. Can you provide unlimited bandwidth with the account?
Pricing for bandwidth charged by data centres in Singapore usually range from SGD200 to SGD700
per Mbps. Each Mbps translate to ~200GB/mth after overheads. Therefore it is not realistic
to provide unlimited bandwidth, unless customers are willing to pay accordingly. With $10/mth
accounts, once you use more than 10GB, we are losing money with each additional GB you use.
Therefore we set a bandwidth quota for all accounts, and make sure we can deliver on what we
8. Other companies provide 99.99% uptime guarantees with money back. Do you provide any
guarantees for your server uptime?
No, we do not. We are probably the only web hosting company in Singapore using 4-CPU Opteron
servers with cPanel (as of Apr 2006). Most companies use 1 or 2-CPU servers with IDE or SATA harddisks. We
simply use the best equipment. When you see companies offering hosting on 4-CPU Opteron servers
in Singapore, they are probably our resellers. Several of our resellers do offer uptime guarantees,
eg. 99.95%, 99.99% etc with money back guarantees to their customers. The guarantees are offered on
OUR equipment and OUR service, with THEIR money. This is how much our resellers trust us. However,
we simply do not offer any guarantee; not to end users, not to resellers. What we do, however,
is to provide the best equipment we can, and the best people to administrate the servers. This
ensures the uptime of our servers are excellent, and one of the best in the industry.
Our system uptime record for our first 4-CPU (Quad) Opterons is currently at 423 days, without
reboot (as of 10 Apr 06). This record is achieved on a production system with over 600 websites.
This is not typical of most servers.
Based on Netcraft.com records of servers in 1-Net data centre, this is the 2nd best record out
of hundreds of servers. The top spot is taken by a mail server operated by 1-Net. It is very
difficult to achieve even 99.9% for a server, even with RAID 1, and 3 x PSUs, and ECC RAM.
That's just for the server hardware alone. The kernel can go into panic, processes can and do
die, memory leaks do occur, and denial of service attacks can bring the network to a crawl.
Our customers are IT professionals. Professionals know excellent uptime is difficult to achieve
consistently, over a long period of time. If Xssist ever promise uptime figures, we will have
multiple servers in multiple datacentres around the world; a proven method of failover, even for
DNS i.e. cached entries on PCs won't defeat the failover; a proven method of replication for your
database, a proven method of storing your mails over a distributed storage.
With all the above in place, we will promise you, not 99.95%, not 99.9%, but 99% uptime. Without
the above, Xssist will not provide any guarantees. When we say uptime, it is total uptime. Scheduled
maintenance included, no excuses. We are just silly about it. Some of our competitors will promise
you anything; 99.995% uptime if you ask for it. We simply will not promise anything, even 99%. Even
when our equipment is 3 to 10 times more expensive than the average competitor.
If we lose potential customers because of this stubborn view on uptime guarantees, it is a loss
for us, but we will bear with it, and wait for the next opportunity to work with you. Xssist will
not lie to get your business.
9. What are the php settings on the server? Is php run as CGI or Apache module? Is ming installed?
Is pgsql available? Do you have ionCube?
Please see the following URLs:
If you need software that is not shown in the above URL, let us know.
- For punggol.sg server
- For sengkang.sg server
10. Is ASP available? Do you provide hosting on windows servers?
No, we specialise in Linux servers. Although ASP can be supported on Linux platform, it is
better to get hosting on a windows server instead.
1. Why do you provide mainly Opteron servers and not Pentium or Xeon servers?
As of Apr 2006, Opteron CPUs uses less electricity, generate less heat and perform better
than Pentium and Xeon CPUs. We like to provide the better CPU for customers. However,
if your preference is for Pentium or Xeon CPUs, we can provide a custom quote for you, with your
preferred brand and model, eg. Dell, HP, IBM or Sun.
2. Can you bundle cPanel, directadmin or ensim with the servers?
Yes, we can bundle the control panels with your dedicated servers. Our specialty is in cPanel.
We are a cPanel Partner NOC.
3. Can I put in a GSM modem for sending out SMS, together with the dedicated server?
Yes, you can.
Virtual Private Servers (VPS)
1. What is the software used for your VPS? Is it Virtuozzo, Xen or UML?
We use Virtuozzo for our VPS servers. This is because Virtuozzo is mature compared to Xen,
and resources are burstable. Virtuozzo also provides web based interface for users to control
their VPS, including backup, restore and restart.
2. What is the server specifications that VPS is on?
We use a Quad Opteron 846, with 8GB RAM, and 15K RPM SCSI harddisks, multiple PSUs, and RAID 1
for the OS.
3. What does dedicated SCSI harddisk mean?
We provide each VPS with its own harddisk. This is much better than the competition where a
few VPS will share a single harddisk. Sharing the same physical harddisk means any VPS can
eat up all the disks IOs for the single shared harddisk, and deprive your VPS of resources.
In future, it may be possible for VPS software to dedicate disk IOs. However, at Xssist,
we solve this problem today, by giving each premium VPS its own harddisk. You do not share
that harddisk with anyone. It is dedicated to your premium VPS only. Not only do we give you a
dedicated harddisk with your VPS, it is a 15K RPM SCSI harddisk, not IDE or SATA.
4. Can the VPS come with RAID 1?
Absolutely. We can provide software RAID1 for your VPS; and additional harddisks if required
for your backups.
5. Can cPanel be bundled with the VPS?
Yes. We can provide cPanel with the VPS. We are an authorized cPanel Partner NOC.
6. What kind of equipment are you using for VPS (Virtuozzo)?
Quad Opteron 850, 16GB RAM, 15K RPM SCSI harddisks.
Photo of server and external SCSI storage before deployment.
cPanel & WHM
Support for cPanel and WHM issues is for Xssist customers only. If you are not our customer,
please contact your reseller for support.
1. Problems encountered with Horde after upgrading cpanel. eg. cannot login
Please try the following steps:
After completion of each step, test if the problem is fixed, before going on.
- WHM -> SQL services -> Repair a database -> horde -> Repair Database
- rm -f /tmp/horde*
- /etc/init.d/cpanel restart
- /scripts/upcp --force
2. Quota information in cPanel does not match the actual usage, and quota given. eg. actual
usage (using du -sk) is 150MB, account quota is 500MB (also shown when quota username is used),
but space left shows as 50MB.
cPanel caches information in ~username/.cpanel-datastore/quota_-v
Try deleting the file, and check if this fixes the problem.
3. chkservd doesn't work after server reboot or crash.
Check for the file /var/run/chkservd.suspend
If it exists, remove it.
4. The disk quota of all the sites on my server seems corrupted. They are all shown as
unlimited on whm except for a few sites which I have just created or edited.
Quota can be reset to package defaults by running
5. I cannot login to webmail. The domain also does not show up on WHM.
Do the following:
If there are error messages regarding conflicting domains, comment out the domains from the
accounts that should not own the domains, in /var/cpanel/users/$username and redo the above
- touch /etc/passwd
6. There is no option for me to show what features is available to resellers.
I created a dummy reseller, but then when logon to whm, under accounts, i don't see any options
Under reseller centre:
- Go to reseller modifications, select the reseller from the dropdown box
- Go to "edit privileges/ nameservers"
- Scroll down to "Features Limits"
- Click on the features you want to enable.
7. Customer can no longer upload files using file manager after I upgraded cPanel/ WHM.
Errors show quota exceeded, and virus detected.
Under WHM -> Tweak Settings, set the following:
The maximum file size in MB allowed for upload through cPanel File manager.
Use "unlimited" for unlimited: (specify something)
The minimum filesystem quota space in MB required after file upload through cPanel File manager
(Default 5MB). This will prevent users from hitting their quota limit through File Manager file
uploads: (specify something)
When you type these two values the cPanel´s uploads through file manager will work.
8. After updating cPanel, edit DNS zone does not show any domains.
Try the following:
- /scripts/perlinstaller Compress::Raw::Zlib
- /scripts/perlinstaller --force Scalar::Util
9. I deleted the dns zone for several domains in cpanel and re-add DNS zone for the domains.
Now the domains cannot be managed through cpanel.
This will show the errors in the /var/cpanel/users/* files.
Edit the files accordingly. Usually, there would be no entry for DNS. If you have a XDNS1 entry
that has the domain name in question, change that entry to DNS. If there is a duplicate entry in
another file, decide which user should own the domain, and edit accordingly.
10. My domains are not resolving. ps -ef|grep named shows that it is reading config files from chroot.
How to fix this?
- edit /etc/sysconfig/named and remove the line ROOTDIR=/var/named/chroot
- /etc/init.d/named restart
11. One of our new clients is having some problems with her Fantastico. When you click on Fantastico,
you get the following error message:
Parse error: syntax error, unexpected $end in
/tmp/cpanel_phpengine.1171296878.313801619.74984760494 on line 105
Please check quota for the account. The disk usage may have reached the quota limit.
12. How to install postgresql 8.x or higher when cpanel installs a lower version?
Note: Installing your own version of postgresql may break cPanel backup of postgresql databases.
If that occurs, you need to do your own backups of postgresql using pg_dumpall
- Use cpanel to install postgresql using /scripts/installpostgres
- download latest rpms from www.postgresql.org
- rpm -e the old version installed by cPanel, except for rh-postgresql-libs
Note: Claimed by a poster on linuxjunkies.in as follows: "Do not remove rh-postgresql-libs.
It is needed to courier-authlib compatibily"
YMMV, so please try with and without rh-postgresql-libs, and feedback on your results.
- run /etc/init.d/postgresql initdb
Note: If you need UTF8 or other encoding, use -E UTF8 etc. otherwise it will default to ascii
- go to WHM, and config postgresql
If you missed out the step to set the default encoding
If you need to change the default encoding, try the following; it works due to createdb using the
encoding of template1 for new databases.
# su - postgres
$ psql template1
UPDATE pg_database SET datallowconn = TRUE where datname = 'template0';
UPDATE pg_database SET datistemplate = FALSE where datname = 'template1';
drop database template1;
create database template1 with template = template0 encoding = 'UNICODE';
UPDATE pg_database SET datistemplate = TRUE where datname = 'template1';
UPDATE pg_database SET datallowconn = FALSE where datname = 'template0';
13. I keep getting unrouteable mail.
Check /var/log/exim_mainlog for the message id of the mails that are getting rejected.
Then grep for the message id in /var/log/exim_mainlog; see where the error is coming from.
A very common problem that causes "unrouteable mail domain" is the "Max Outgoing email per domain per hour"
setting. If you set 200 mails per hour per domain, once the quota is reached, i.e. the domain sent
out 200 mails per hour, that domain will start getting "R=fail_remote_domains: unrouteable mail domain"
errors. Just set a higher figure, and the problem goes away. Of course, please check that the domain is
not being used to send out spam.
SPAM and SPF
Mail spamming (any unsolicited mailing) is not allowed in any of our servers or colocation or
VPS or dedicated servers. Any account or server used to send spam will be suspended.
1. Over the past couple of weeks. I have been receiving a bunch of emails saying delivered mailed was returned,
blocked, also have some email send to our email account that does not exit, etc. Obviously, I have NOT sent mail
to any of these returned/blocked addresses so my question is...
Could someone be "hacking" me and somehow be using my email address to send out spam? Or are these emails
actually spam TO me. Many of these contain attachments.
I'm even receiving some mail FROM "email@example.com"
I have run SpamAssassin in cpanel. I also have put these email address to Blacklist, but still have some email come to our email server,
If I am being "hacked" what do I do?
You are not hacked. What happens is the people doing spamming uses your domain for some of the emails that are going out.
To reduce this, you can use SPF, http://en.wikipedia.org/wiki/Sender_Policy_Framework
These are rules that can be set in the DNS information for your domain. It defines which IP are supposed to send mails from @yourdomain.com
It will reduce but not totally eliminate these spammers from faking mails using your domain. The reduction comes about because
the receipients mail servers which follow SPF will reject the mails, and spammers want their mails to be delivered.
So if your domain has SPF defined, spammers will not want to use your domain name.
1. How do I switch to your email hosting from my present service provider?
Will there be any downtime? And, do you have web based email manager program and
Yes, control panel is provided for you to manage email accounts,
eg. add, delete, change password, set disk quota for each email account
Web-based email is also available:
Horde and Squirrelmail is provided. Others can be installed by request.
In addition, you can use mail client such as outlook to retrieve mails using POP or IMAP
The process for migration is as follows:
- I will setup Xssist name servers to have records for yourdomain.com
I point the domain MX (mail) entries to your current mail server.
- You need to change your whois records to switch the nameservers for the domain to Xssist
- DNS propagation will take place across the Internet. Both your former and Xssist name
servers may be used by other email servers to check where to send mails to. However, both former
and Xssist name servers will ask mail servers to send mails for yourdomain.com to your former
At this point, you still read and send mails from your former mail servers.
- We will wait a few days after the DNS changes.
- I will switch yourdomain.com to use Xssist mail server.
- On the next working day, your staff should download all remaining mails from your current
mail server. Then switch to using Xssist webmail/ mail server.
- Other than migrating the mail server, you need to prepare your staff to use a different
webmail, and change configuration on their PCs to download from a different mail server.
- At this stage, all your mails for yourdomain.com should be received and sent from Xssist
Troubleshooting for dedicated servers
1. Apache refuses to start, and I get the follow error message in apache logs
semget: No space left on device
There is a semaphore leak. i.e. semaphores are used but not released. Clear it by
enabling login for the user "nobody", and su - nobody ipcclean ; this assumes you are
running apache as "nobody". Before doing so, you can check this is the problem by using ipcs.
You can delete each semaphore using ipcrm
2. Why do I get connection refused for rsync?
#rsync rpm/ firstname.lastname@example.org:.
connect to address 126.96.36.199 port 544: Connection refused
Trying krb4 rsh...
connect to address 188.8.131.52 port 544: Connection refused
trying normal rsh (/usr/bin/rsh)
184.108.40.206: Connection refused
rsync: connection unexpectedly closed (0 bytes read so far)
rsync error: error in rsync protocol data stream (code 12) at io.c(170)
Usually servers are no longer (for a number of years) setup to use rsh. rsync defaults to rsh. use the RSYNC_RSH variable to change to ssh,
eg. export RSYNC_RSH=ssh
then run rsync
This will also fix a problem with vzmigrate for migrating parallels (formerly virtuozzo) Containers (or VEs) from one host to another. If you
get an error that says something like "broken pipe", check if its due to rsync using rsh and not being able to connect to the destination server.